Thomson Newspapers Co. v. Canada (Attorney General), [1998] 1 S.C.R. 877

Thomson Newspapers Company Limited, doing business as

The Globe and Mail, The Evening Telegram, Winnipeg

Free Press and Times‑Colonist, and Southam Inc. Appellants

v.

The Attorney General of Canada Respondent

and

The Attorney General of British Columbia

and the Canadian Civil Liberties Association Interveners

Indexed as: Thomson Newspapers Co. v. Canada (Attorney General)

File No.: 25593.

1997: October 9; 1998: May 29.

Present: Lamer C.J. and L’Heureux‑Dubé, Sopinka,* Gonthier, Cory, McLachlin, Iacobucci, Major and Bastarache JJ.

on appeal from the court of appeal for ontario

Constitutional law ‑‑ Charter of Rights ‑‑ Freedom of expression ‑‑ Opinion surveys ‑‑ Elections ‑‑ Federal elections legislation prohibiting publication, dissemination or broadcasting of opinion survey results on last weekend of election campaign and on polling day ‑‑ Whether legislation infringes freedom of expression ‑‑ If so, whether infringement justifiable ‑‑ Canadian Charter of Rights and Freedoms, ss. 1, 2(b) ‑‑ Canada Elections Act, R.S.C., 1985, c. E-2, s. 322.1.

Constitutional law ‑‑ Charter of Rights ‑‑ Right to vote ‑‑ Access to information during election ‑‑ Restrictions on opinion survey results ‑‑ Canadian Charter of Rights and Freedoms, s. 3 ‑‑ Canada Elections Act, R.S.C., 1985, c. E‑2, s. 322.1.

The appellants brought an application for a declaration that s. 322.1 of the Canada Elections Act violates freedom of expression and the right to vote guaranteed by ss. 2(b) and 3 of the Canadian Charter of Rights and Freedoms. The impugned section prohibits the broadcasting, publication or dissemination of opinion survey results during the final three days of a federal election campaign. The Ontario Court (General Division) denied the appellants’ application, holding that s. 322.1 did not violate a citizen’s right to vote and that, although the section infringed freedom of expression, it was justified under s. 1 of the Charter. The Court of Appeal affirmed the judgment.

Held (Lamer C.J. and L’Heureux-Dubé and Gonthier JJ. dissenting): The appeal should be allowed.

Per Cory, McLachlin, Iacobucci, Major and Bastarache JJ.: Section 322.1 of the Canada Elections Act, which applies only to “new” poll results, infringes s. 2(b) of the Charter. The publication of opinion survey results is an activity that conveys meaning and therefore falls within the ambit of s. 2(b). By prohibiting the broadcasting, publication or dissemination of opinion survey results during the final three days of an election campaign, s. 322.1 restricts freedom of expression.

Section 322.1 is not justified under s. 1 of the Charter. All the steps of a s. 1 analysis must be undertaken with a close attention to context. Characterizing the context of the impugned provision is important in order to determine the type of proof which a court can demand of the legislator to justify its measures under s. 1. In the course of a contextual approach under s. 1, the vulnerability of the group which the legislator seeks to protect, that group’s own subjective fears or apprehension of harm, the inability to measure scientifically a particular harm, and the efficaciousness of a remedy, are all factors which the court must take into account in assessing whether a limit has been demonstrably justified according to the civil standard of proof. Another factor to be considered is the nature of the activity which is infringed. The degree of constitutional protection may vary depending on the nature of the expression at issue. Here, the speech infringed is political information. Opinion surveys regarding political candidates or electoral issues are part of the political process and, thus, at the core of expression guaranteed by the Charter. The nature of the expression at issue suggests that a deferential approach is inappropriate in this case.

While the objective of providing a period of rest and reflection for voters prior to going to the polls is not a pressing and substantial objective, the objective of guarding against the possible influence of inaccurate polls late in the election campaign by allowing for a period of criticism and scrutiny immediately prior to election day is of sufficient importance to meet the first step of the s. 1 analysis. The purpose of this particular limitation on expression is to ensure that information which the evidence indicates has an important influence on the choice of at least some voters is presented according to the standards of accuracy which polls are normally expected to attain. To the extent that the votes of some might be distorted as a result of polls being presented in a misleading fashion, such a distortion is clearly a matter which the government may legitimately be concerned to remedy.

The three-day blackout period on the publication of polls will serve, to some degree, the purpose of preventing the use of inaccurate polls by voters by giving critics the opportunity to assess the methodological information made available by the pollster and to question the validity of the poll on that basis. To that extent, the ban is rationally connected to the purpose of the legislation. However, since s. 322.1 does not require the publication of methodological information, the most that could be achieved by the blackout period, if an opinion survey is released without such information, is that the validity of a poll could be undermined by pointing out the failure of the pollster to publish the methodology of the poll.

Section 322.1 does not minimally impair freedom of expression. The section is a very crude instrument in serving the government’s purpose. The social science evidence did not establish that Canadian voters are a vulnerable group relative to pollsters and the media who publish polls. The presumption should be that the Canadian voter is a rational actor who can learn from experience and make independent judgments about the value of particular sources of electoral information. While some voters clearly do consider polls to be of some value in making their electoral decision, no evidence has been adduced that voters have suffered from any misapprehensions regarding the accuracy of any single poll. Voters are constantly exposed to opinion poll results throughout the election and a single inaccurate poll result is likely to be spotted and discounted appropriately. This is not an appropriate case for the government to respond to the paucity of evidence by relying on the “reasoned apprehension of harm” test. First, the claims of widespread or significant harm based on a logical inference derived from surrounding factors are not compelling in the context of factors which, as in this case, refute such logical inferences. Second, the government is not dealing with a vulnerable group which is in danger of manipulation or abuse by the pollsters or the media because of an essential opposition of interests, or because of the nature of the speech itself. Nor is there a shared understanding amongst Canadians that a single inaccurate poll will mislead them to any undue extent.

Where the contextual factors indicate that the government has not established that the harm which it is seeking to prevent is widespread or significant, a deferential approach to the particular means chosen by the legislature to implement the legislative purpose is not warranted. In this case, s. 322.1 is not narrowly tailored to its objective. The ban is overbroad because it prohibits in the final three days of an election campaign the publication and use by voters of all those polls which would meet the usual standards of accuracy. The ban is also underbroad because it may not adequately disabuse voters of an erroneous impression left by a poll which did not disclose its methodology to critics or the public. The obvious alternative was a mandatory disclosure of methodological information without a publication ban. Although such a provision would still leave the door open to inaccurate poll results published immediately prior to the election having some impact, that possibility would be significantly reduced both by virtue of the reader’s initial access to those methodological data, and by the opportunity for rapid response by parties whose interests are prejudiced by the inaccurate poll. The failure to address or explain the reason for not adopting a significantly less intrusive measure which appears as effective as that actually adopted weighs heavily against the justifiability of s. 322.1. Finally, the experience of the international community is inconclusive.

The doubtful benefits of the ban are outweighed by its deleterious effects. The impact of s. 322.1 on freedom of expression is profound. The section imposes a complete ban on political information at a crucial time in the electoral process. The ban interferes with the rights of voters who want access to the most timely polling information available, and with the rights of the media and pollsters who want to provide it. Although it is conceivable that some indeterminate number of voters might be unable to spot an inaccurate poll result and might rely to a significant degree on the error, thus perverting their electoral choice, the government cannot take the most uninformed and naive voter as the standard by which constitutionality is assessed. A measure which decides that information which is desired and can be rationally and properly assessed by the vast majority of the voting electorate should be withheld because of a concern that a very few voters might be so confounded that they would cast their vote for a candidate whom they would not have otherwise preferred cannot be accepted. Given the state of the evidence adduced on this issue, the postulated harm will seldom occur. The benefits of the ban are, therefore, marginal. The deleterious effects, however, are substantial. The ban sends the general message that the media can be constrained by government not to publish factual information. As well, the ban interferes with the media’s reporting function with respect to the election. Further, by denying access to electoral information which some voters may consider useful, the ban interferes not only with their freedom of expression, but also with their perception of the freeness and validity of their vote. In sum, the very serious invasion of the freedom of expression of all Canadians is not outweighed by the speculative and marginal benefits postulated by the government.

In light of the conclusion that s. 322.1 of the Canada Elections Act is an unjustified limit on free expression, it is unnecessary to determine whether the section constitutes an infringement of the right to vote protected by s. 3 of the Charter.

Per Lamer C.J. and L’Heureux-Dubé and Gonthier JJ. (dissenting): Section 322.1 of the Canada Elections Act does not infringe s. 3 of the Charter. A restriction on information would constitute an infringement of the right to vote under s. 3 only if it undermines the guarantee of effective representation. In the instant case, the short blackout period has no such effect. On the contrary, such a period assists effective representation.

While s. 322.1 limits freedom of expression within the meaning of s. 2(b) of the Charter, it constitutes a reasonable limit demonstrably justified in a free and democratic society under s. 1 of the Charter. The objective of preventing the potentially distorting effect of public opinion survey results that are released late in an election campaign when there is no longer a sufficient opportunity to respond is a sufficiently important objective which meets the first step of the s.1 analysis. Opinion polls on election issues influence voters’ decisions and it is important that the information the polls convey not be misleading or inaccurate. By providing for timely publication of poll results to allow scrutiny and criticism, s. 322.1 improves information to the public during election campaigns, enhances the electoral process and strikes a balance between the right to vote and freedom of expression. The social science studies which composed much of the evidence show that there exists a long‑standing concern about the publication of opinion survey results during election campaigns in Canada, including the problems associated with the undue influence, late publication and accuracy of polls. In enacting s. 322.1, Parliament has responded to that concern. The Charter should not become an impediment to social and democratic progress and be made to serve substantial commercial interests in publishing opinion poll results, by defeating a reasonable attempt by Parliament to allay potential distortion of voter choice. Several studies and reports in the last 30 years on the publication of opinion survey results during election campaigns in Canada, as well as bills in the House of Commons, and legislation in other democratic countries, support Parliament’s reasonable finding that the concern at bar was serious.

The second step of the s. 1 analysis ‑‑ the proportionality test‑‑ is also met. First, the rational connection in this case is self‑evident. Opinion polls significantly influence voter choice and electoral campaigns. It follows that the publication of inaccurate, though authoritative, opinion survey results that go uncorrected may well lead to voters making misinformed decisions. Logically, there is a reasoned apprehension that voters will be deprived of the full exercise of their franchise. Its importance is measured by the significant influence of polls on voters and the prevalence of misleading polls. Ensuring that polls that cannot be adequately, publicly and independently evaluated as to their correctness because of insufficient time are not published clearly addresses this problem.

Second, s. 322.1 passes the minimal impairment analysis. Section 322.1 constitutes a genuine mediation between the rights of voters to receive information in a timely fashion, and the right of pollsters and publishers freely to provide the information they want. Not only does the legislation protect the rights of voters but it does so by serving one of the very purposes of freedom of expression ‑‑ informing while allowing for political debate and discussion ‑‑ and by striking a balance between two basic aspects of voters’ right to information ‑‑ the availability of accessible, unrestricted information and the timely availability of factual information that may be misleading so as to allow for scrutiny and criticism. In matching means to ends and asking whether rights are impaired as little as possible, a legislature mediating between the claims of competing groups is forced to strike a balance without the benefit of absolute certainty concerning how that balance is best struck. This Court should not second-guess the wisdom of a legislature in its endeavour to draw the line between competing credible evidence, once it has been established, on the civil standard of proof, that Parliament’s objective was pressing and substantial. At this stage, the question is whether there is a reasonable basis, on the evidence tendered, for concluding that a blackout on all opinion polls during the last weekend of an election campaign and during election day impairs freedom of expression as little as possible given the government’s pressing and substantial objective. Parliament is not bound to find the least intrusive nor the best means. This would be too high a standard for our elected representatives to meet. Here, although one can conceive of alternatives to the impugned measure, there is simply no equally effective alternative to the current short-term blackout for achieving the legislative objective. Based upon the current legislation and the reports and studies available, Parliament reasonably determined that a 72-hour period was necessary to allow meaningful scrutiny of poll results during an election campaign. The 72-hour blackout period is very shortlived and only affects one mode of expression which is not a primary source of information concerning relevant political facts. It mainly constitutes information as to the effect of relevant political information on potential voters. The scope of s. 322.1 prohibits polls of all kinds regardless of their scientific nature or quality because there is no clear cut line between reliable poll results and misleading poll results. Section 322.1 does not apply, however, to the discussion of previously released poll results. The legislation prohibits broadcasting, publication and dissemination and these expressions refer only to the initial release of poll results. Finally, both the motions judge and the Court of Appeal held that inaccurate polls at the end of an election campaign constitute a reasonable concern. Everyone is vulnerable to misinformation which cannot be verified. Our democracy, and its electoral process, finds its strength in the vote of each and every citizen. Each citizen, no matter how politically knowledgeable one may be, has his or her own reasons to vote for a particular candidate and the value of any of these reasons should not be undermined by misinformation. When Parliament identifies one matter of concern, it has no absolute duty to identify and regulate each and every factor. The government does not have to show that this concern is more serious or is causing more harm to the electoral process or to individual voters than any other potentially misleading information. There is no such standard under the Charter.

Third, the salutary effects of s. 322.1 concerning both the right to vote and freedom of expression outweigh its deleterious effects. The salutary effect of s. 322.1 is to promote the right of voters not to be misled in the exercise of their right to vote. Section 322.1, however, deprives some voters, who rely on polls to make their decision, of late campaign opinion poll results. This deleterious effect is quite limited when one considers the delay between conducting the poll and ultimately publishing its results. As to the effects of the measure on freedom of expression, s. 322.1 has a positive impact, promoting debate and truth in political discussion since it gives voters the opportunity to be informed about the existence of misleading factual information. Although s. 322.1 precludes the media from publishing polls on the last weekend of the election campaign and on polling day, this ban causes only minimal impairment to freedom of expression because of its very short duration and because of the lack of satisfactory alternatives available to tailor the measure to the legislative objective.

Cases Cited

By Bastarache J.

Distinguished: R. v. Butler, 1992 124 (SCC), [1992] 1 S.C.R. 452; Libman v. Quebec (Attorney General), 1997 326 (SCC), [1997] 3 S.C.R. 569; Irwin Toy Ltd. v. Quebec (Attorney General), 1989 87 (SCC), [1989] 1 S.C.R. 927; RJR‑MacDonald Inc. v. Canada (Attorney General), 1995 64 (SCC), [1995] 3 S.C.R. 199; R. v. Keegstra, 1990 24 (SCC), [1990] 3 S.C.R. 697; referred to: Dagenais v. Canadian Broadcasting Corp., 1994 39 (SCC), [1994] 3 S.C.R. 835; Reference re Provincial Electoral Boundaries (Sask.), 1991 61 (SCC), [1991] 2 S.C.R. 158; Sauvé v. Canada (Attorney General), 1993 92 (SCC), [1993] 2 S.C.R. 438; Haig v. Canada, 1993 58 (SCC), [1993] 2 S.C.R. 995; Eur. Court H.R., Mathieu‑Mohin and Clerfayt case, judgment of 2 March 1987, Series A No. 113; Bowman v. United Kingdom (1996), 22 E.H.R.R. C.D. 13; R. v. Oakes, 1986 46 (SCC), [1986] 1 S.C.R. 103; Edmonton Journal v. Alberta (Attorney General), 1989 20 (SCC), [1989] 2 S.C.R. 1326; Rocket v. Royal College of Dental Surgeons of Ontario, 1990 121 (SCC), [1990] 2 S.C.R. 232; Ross v. New Brunswick School District No. 15, 1996 237 (SCC), [1996] 1 S.C.R. 825; Towne Cinema Theatres Ltd. v. The Queen, 1985 75 (SCC), [1985] 1 S.C.R. 494; Vriend v. Alberta, 1998 816 (SCC), [1998] 1 S.C.R. 493; Harvey v. New Brunswick (Attorney General), 1996 163 (SCC), [1996] 2 S.C.R. 876; R. v. Laba, 1994 41 (SCC), [1994] 3 S.C.R. 965; Butler v. Michigan, 352 U.S. 380 (1957).

By Gonthier J. (dissenting)

Re C.F.R.B. and Attorney‑General for Canada, 1973 788 (ON CA), [1973] 3 O.R. 819; Reference re Provincial Electoral Boundaries (Sask.), 1991 61 (SCC), [1991] 2 S.C.R. 158; R. v. Keegstra, 1990 24 (SCC), [1990] 3 S.C.R. 697; R. v. Oakes, 1986 46 (SCC), [1986] 1 S.C.R. 103; Libman v. Quebec (Attorney General), 1997 326 (SCC), [1997] 3 S.C.R. 569; RJR‑MacDonald Inc. v. Canada (Attorney General), 1995 64 (SCC), [1995] 3 S.C.R. 199; Irwin Toy Ltd. v. Quebec (Attorney General), 1989 87 (SCC), [1989] 1 S.C.R. 927; Ross v. New Brunswick School District No. 15, 1996 237 (SCC), [1996] 1 S.C.R. 825; R. v. Zundel, 1992 75 (SCC), [1992] 2 S.C.R. 731; Ford v. Quebec (Attorney General), 1988 19 (SCC), [1988] 2 S.C.R. 712; RWDSU v. Dolphin Delivery Ltd., 1986 5 (SCC), [1986] 2 S.C.R. 573; R. v. Butler, 1992 124 (SCC), [1992] 1 S.C.R. 452; Canada (Attorney General) v. Somerville, 1996 ABCA 217, [1996] 8 W.W.R. 199; Harvey v. New Brunswick (Attorney General), 1996 163 (SCC), [1996] 2 S.C.R. 876; R. v. Edwards Books and Art Ltd., 1986 12 (SCC), [1986] 2 S.C.R. 713; McKinney v. University of Guelph, 1990 60 (SCC), [1990] 3 S.C.R. 229; Dagenais v. Canadian Broadcasting Corp., 1994 39 (SCC), [1994] 3 S.C.R. 835.

Statutes and Regulations Cited

Canada Elections Act, R.S.C., 1985, c. E‑2, ss. 213(1) [rep. & sub. 1993, c. 19, s. 106], 255, 256, 322.1 [en. idem, s. 125].

Canadian Charter of Rights and Freedoms, ss. 1, 2(b), 3, 33.

Criminal Code, R.S.C., 1985, c. C‑46, s. 126(1).

Constitution Act, 1982, s. 52.

Convention for the Protection of Human Rights and Fundamental Freedoms, First Protocol, Europ. T.S. No. 9, Art. 3.

Election Act, R.S.B.C. 1996, c. 106, s. 235.

Ontario Rules of Civil Procedure, R.R.O. 1990, Reg. 194, rule 14.05(3)(g.1).

Authors Cited

Ansolabehere, Stephen, and Shanto Iyengar. “Of Horseshoes and Horse Races: Experimental Studies of the Impact of Poll Results on Electoral Behavior”, Political Communication, vol. 11, No. 4, 1994, 413-430.

Cameron, Jamie. “The Past, Present, and Future of Expressive Freedom Under the Charter” (1997), 35 Osgoode Hall L.J. 1.

Canada. Committee on Election Expenses. Report of the Committee on Election Expenses. Ottawa: Queen’s Printer, 1966.

Canada. Privy Council. White Paper on Election Law Reform. Ottawa, 1986.

Canada. Royal Commission on Electoral Reform and Party Financing. Reforming Electoral Democracy: Final Report, vol. 1. Ottawa: Minister of Supply and Services Canada, 1991.

Canada. Senate. Debates of the Senate, April 29, 1993, p. 3117.

Canada. Senate. Standing Committee on Legal and Constitutional Affairs. Proceedings, Issue No. 41, May 6, 1993, pp. 41:14‑41:15, 41:17‑41:18.

Feasby, Colin C. J. “Public Opinion Poll Restrictions, Elections, and the Charter” (1997), 55(2) U.T. Fac. L. Rev. 241.

Herbst, Susan. Numbered Voices: How Opinion Polling Has Shaped American Politics. Chicago: University of Chicago Press, 1993.

Hogg, Peter W. Constitutional Law of Canada, vol. 2, loose‑leaf ed. Scarborough, Ont.: Carswell, 1992 (updated 1997, release 2).

Hoy, Claire. Margin of Error: Pollsters and the Manipulation of Canadian Politics. Toronto: Key Porter Books, 1989.

Johnston, Richard, et al. Letting the People Decide: Dynamics of a Canadian Election. Montreal: McGill‑Queen’s University Press, 1992.

Lachapelle, Guy. Polls and the Media in Canadian Elections: Taking the Pulse. Ottawa: Royal Commission on Electoral Reform and Party Financing, 1991.

McAllister, Ian, and Donley T. Studlar. “Bandwagon, Underdog, or Projection? Opinion Polls and Electoral Choice in Britain, 1979‑1987”, Journal of Politics, vol. 53, No. 3, August 1991, 720-41.

APPEAL from a judgment of the Ontario Court of Appeal (1996), 1996 890 (ON CA), 30 O.R. (3d) 350, 92 O.A.C. 290, 138 D.L.R. (4th) 1, 37 C.R.R. (2d) 225, [1996] O.J. No. 2829 (QL), dismissing the appellants’ appeal from a judgment of the Ontario Court (General Division) (1995), 1995 7062 (ON SC), 24 O.R. (3d) 109, [1995] O.J. No. 1375 (QL), dismissing an application for a declaration that s. 322.1 of the Canada Elections Act is unconstitutional. Appeal allowed, Lamer C.J. and L’Heureux-Dubé and Gonthier JJ. dissenting.

W. Ian C. Binnie, Q.C., and Michael J. Bryant, for the appellants.

Roslyn J. Levine, Q.C., and Gail Sinclair, for the respondent.

Joseph J. Arvay, Q.C., for the intervener the Attorney General of British Columbia.

Sydney L. Goldenberg and Stephen L. McCammon, for the intervener the Canadian Civil Liberties Association.

The reasons of Lamer C.J. and L’Heureux-Dubé and Gonthier JJ. were delivered by

1 Gonthier J. (dissenting) -- I have had the benefit of the reasons for judgment of my colleague, Justice Bastarache. I refer to his summary of the judgments below and statement of the issues. While I agree with his statutory interpretation of s. 322.1 of the Canada Elections Act, R.S.C., 1985, c. E-2, with his views as to the scope of s. 3 of the Canadian Charter of Rights and Freedoms, and with his position that there is an infringement of freedom of expression, within the meaning of s. 2(b) of the Charter, I am not in agreement with his reasons and his disposition of the appeal with respect to the justification of the infringement. In my view, the said infringement is justified under s. 1 of the Charter.

I. Introduction and Historical Background

2 In recent political history, polls have had a substantial impact on the strategies of candidates and the policies of governments. They have become a permanent feature of Canadian politics. It is said that polls tend to reduce the level of political discourse to the lowest common denominator: principles are sacrificed for percentage points (C. C. J. Feasby, “Public Opinion Poll Restrictions, Elections, and the Charter” (1997), 55(2) U.T. Fac. L. Rev. 241, at p. 244). They tend to preempt the discussion of issues and short-circuit the democratic process:

Conversation, as theorists from Tarde to Habermas have argued, is fundamental to the construction of a democratic public sphere, and polls do not seem to generate interpersonal communication. In a way, polls make many political discussions superfluous, since they give the illusion that the public has already spoken in a definitive manner. [Emphasis in original.]

(S. Herbst, Numbered Voices: How Opinion Polling Has Shaped American Politics (1993), at p. 166.)

To the extent that media coverage of election campaigns focuses more on polling results, it focuses less on the merits of the candidates and their positions and tends to distract voters’ attention from substantive issues pertaining to the good government of the country. The reliance on polls has become so pervasive that some commentators characterize election campaigns as “horse races” (S. Ansolabehere and S. Iyengar, “Of Horseshoes and Horse Races: Experimental Studies of the Impact of Poll Results on Electoral Behavior”, Political Communication, vol. 11, No. 4, 1994, 413-430). The following quotation strikingly illustrates the developing situation:

The 1980s is a decade in which media polling has become an epidemic; in which polling has become not just a political tool, an early-warning guidance system, but an occasional substitute for policy itself; in which democracy itself was undermined by the pervasive influence of polls and pollsters who, for all practical purposes, replaced elected representatives, including cabinet ministers, and traditional political strategists as the major determinants of political action. On the strength of their computer printouts, pollsters came to be viewed as sainted public oracles, supposedly capable not only of probing our innermost thoughts and feelings, but of predicting our future actions as well.

Polls have become a perplexing reality in the political process; still a legitimate, albeit imperfect, measure of public attitudes, they are afforded an unhealthy level of credence in the affairs of state by political players and the public alike.

(C. Hoy, Margin of Error: Pollsters and the Manipulation of Canadian Politics (1989), at pp. 39-40.)

The problem becomes more acute when some voters consciously use survey results to make decisions, despite the fact that some polls may be inaccurately conducted, misrepresented by the media or misunderstood by the public. The accuracy of opinion poll results may be deceptive and the credibility owed to them exaggerated. The closer the polling day, the less time there is to assess, scrutinize, and possibly correct polls. While the “horse race” turn that election campaigns take is unfortunate, voters have the right to choose the information that they want to rely on in deciding how to vote. If a voter wants to vote strategically, he may rely on poll results to make his decision. However, he will not be well served by an inaccurate or misleading poll. The expression he tries to convey through his vote could be unsubstantiated or misinformed.

3 It should be noted that overall results may not reveal the actual impact of opinion polls on individual voters. For instance, the choice of a voter who wants to support the leading candidate may be cancelled out by the vote of someone who wants to support the trailing candidate, while both were misled by the same poll. Our democracy, and its electoral process, finds its strength in the vote of each and every citizen. Our Charter provides that each citizen has the right to vote. One should not find solace in the thought that two mistakenly cast votes may perchance cancel each other out.

4 To quote the finding of Somers J., the motions judge below, “[t]he effect of polling has been a long-standing concern amongst those involved in the study of elections and with those who actively partake in the elections themselves” ((1995), 1995 7062 (ON SC), 24 O.R. (3d) 109, at p. 121). In 1966, a government-appointed Committee on Election Expenses the Barbeau Committee in the course of its work on the limitation of election expenses, identified opinion polls as an area of concern. In its report, at p. 51, the Committee recommended that the publication of opinion surveys be completely banned throughout election campaigns as it considered “their uncontrolled use for public purposes improper”. The Government recognized that this was an issue of some concern in its White Paper on Election Law Reform in 1986. The White Paper pointed out that up to that time, more than 20 private members’ bills had been introduced in the House of Commons with the purpose of either prohibiting the publication of opinion surveys or controlling the methodology of such surveys published during campaigns. Although none of these bills were passed, they were reflective of the public concern over this issue, since the sponsors of the bills represented different regions of the country, as well as the various political parties which have been represented in the House of Commons over the past 30 years. The Government argued that certain precautions should be taken to protect the public against abuses in this area. In so doing, it recommended that methodological information be included in the publication or broadcast of opinion surveys during an election campaign.

5 By an Order in Council dated November 15, 1989, a Royal Commission on Electoral Reform and Party Financing, chaired by Mr. Pierre Lortie (the “Lortie Commission”), was appointed to inquire into and report on the appropriate principles and process that should govern the election of members of the House of Commons and the financing of political parties and of candidates’ campaigns. It held hearings across Canada and considered the issue of opinion surveys in some depth. During the hearings, at least 90 briefs were submitted regarding the influence of public opinion surveys. Seventy percent of these briefs favoured government regulation of such polls during elections.

6 The Commission retained Professor Guy Lachapelle, Associate Professor in the Department of Political Science at Concordia University, to research and study this issue. His research study, entitled Polls and the Media in Canadian Elections: Taking the Pulse (“Lachapelle Study”), was published in 1991. Somers J. found the Lachapelle Study, “because of its breadth and thoroughness to be the most credible evidence put before the court” (p. 116). Among other things, Lachapelle surveyed the briefs filed with the Commission concerning opinion polls. At para. 8 of his affidavit filed in the instant case (Case on Appeal, at pp. 92 et seq.), Professor Lachapelle stated:

The basic argument of most of these briefs was that polls have undue influence on elections, especially as a potential influence on voters. However, no definitive conclusion about the actual impact of public opinion surveys could be drawn from these briefs. [Emphasis added.]

Professor Lachapelle admitted that the actual impact of public opinion surveys was, and continues to be, a matter of controversy even among researchers. He went on to say:

Therefore, the academic literature is highly divided on the impact of public opinion surveys during and outside election periods. Further, this literature cannot demonstrate that polls do, in fact, have a major impact on the outcome of elections. Even the concept of strategic voting cannot be scientifically proven. However, we cannot reject or ignore the hypothesis that an impact does exist, and has consequences for voting behaviour. [Emphasis added.]

(Lachapelle’s affidavit, at para. 21.)

The foregoing debate appears centered on the effect of polls on the outcome of elections. Of greater import in the case at bar is the effect of polls on individual voters. The Lortie Commission Final Report (Reforming Electoral Democracy (1991), vol. 1) found, at p. 455, that “the publication of opinion polls during election campaigns is controversial”. However, it reached the conclusion that “[n]otwithstanding the frequent assertion of pollsters that their data have minimal influence on voters, recent research provides strong support for the proposition that published opinion polls can significantly influence campaigns and voters” (p. 457). It also found that “[r]ecent Canadian research supports the conclusion that published campaign opinion polls create the conditions for a ‘politics of expectations’ that includes both strategic voting and bandwagon effects” (p. 458). “In short, the argument that published polls do not influence voter choice or affect the conduct of campaigns is simply untenable” (p. 458). At para. 18 of his affidavit, Professor Lachapelle mentioned that research on the 1988 federal election shows that the behaviour of some voters is guided partly, but not solely, by the published and broadcast results of public opinion surveys.

7 In his study, Professor Lachapelle mentioned several potential effects of opinion polls on voters at election time:

· the bandwagon effect (electors rally to support the candidate leading in the polls);

· the underdog effect (electors rally to support the candidate trailing in the polls);

· the demotivating effect (electors abstain from voting out of certainty that their candidate will win);

· the motivating effect (electors vote because the polls alert them to the fact that an election is going on);

· the strategic effect (electors decide how to vote on the basis of the relative popularity of the parties according to the polls); and

· the free-will effect (electors vote to prove the polls wrong).

(Lachapelle Study, supra, at pp. 13-14.)

The bandwagon effect is the best documented result of exposure to poll news. Studies found that, while the bandwagon effect may occur in response to any poll, elections provide the most fertile ground for the growth of poll-driven opinion. It is said that the bandwagon effect was observed in British and American elections, and in Quebec during the 1988 federal election (I. McAllister and D. T. Studlar, “Bandwagon, Underdog, or Projection? Opinion Polls and Electoral Choice in Britain, 1979-1987”, Journal of Politics, vol. 53, No. 3, August 1991, 720-41, at p. 736; Ansolabehere and Iyengar, supra, at p. 427; R. Johnston et al., Letting the People Decide: Dynamics of a Canadian Election (1992), at p. 200).

8 In the course of his work, Lachapelle studied the brief of the Quebec Federation of Professional Journalists to the Lortie Commission. In its brief, the Federation conceded that a 48-hour blackout prohibiting all publication or broadcast of opinion polls before the vote would be an acceptable compromise and would respect every individual’s right to have time to reply (Lachapelle Study, supra, at p. 27).

9 At the end of his study, Lachapelle concluded that “there are significant shortcomings in the media treatment of polls” (Lachapelle Study, at p. 154). He listed several factors which could seriously impact on the accuracy and reliability of opinion poll results:

1. The Order and Wording of Questions Asked: For instance, according to Claire Hoy’s statement in 1989 (Hoy, supra), Angus Reid differs from the other polling organizations in that it asks which of the three leaders is the most popular before asking which of the parties the respondent supports. In addition, Reid asks the question on voting intentions at the end of the questionnaire, whereas Environics, for example, asks it at the beginning (Lachapelle Study, at pp. 90 and 95);

2. The Different Methods of Distribution of the Undecided Voters: In the 1988 federal election, no polling organization made a distinction between respondents who intended to refrain from voting, intended to spoil their ballot, did not know how they were going to vote, or refused to answer, although the first two instances are hardly cases of indecision; they express a clear opinion. The treatment of undecided respondents varies considerably and may actually misrepresent reality (Lachapelle Study, at pp. 95-96);

3. Data-Collection Methods: During the 1988 federal election campaign, the polling organization Gallup Inc. conducted its surveys both by telephone and in face-to-face interviews, separately or together. This poses numerous problems, especially when the results are presented together without comparing telephone results with those obtained through personal interviews (Lachapelle Study, at p. 97). It is difficult for the elector to grasp all the methodological nuances of sampling (Lachapelle Study, at p. 117). The reader requires a certain attentiveness to recognize that Gallup’s methodology varied from one poll to the next, although the press reports contain the information (Lachapelle Study, at p. 116);

4. Margin of Error: The margin of error for a particular question may be higher than the average margin because not all the interviewees respond to all the questions. This is not expressly mentioned in the polls (Lachapelle Study, at p. 104);

5. Timeframe for the Interviews: For a cross-Canada survey, some organizations take up to seven days to interview their respondents, which may reduce the accuracy and reliability of the “snapshot” (Lachapelle Study, at pp. 113-15 and 116);

The last poll of the 1988 campaign, broadcast two days before election day, did not mention the dates of the interviews. As the study put it: “How could even moderately perspicacious citizens be expected to react and exercise their right of rejoinder? Given the amount of time left, this right was rendered ineffective.” (Lachapelle Study, at p. 116.)

10 Predicting regional voting intentions from a cross-Canada survey is also subject to considerable risk, though it became widespread in recent elections. The validity of certain local polls can also be questioned when the interview periods extend over several weeks if not months (Lachapelle Study, at p. 145).

11 In spite of their serious impact on the reliability of polls, these factors, just to name a few, are often overlooked by the public when considering poll results but are carefully taken into account by analysts.

12 On the basis of his analysis of the domestic and international context, Professor Lachapelle made the following recommendations to the Lortie Commission:

(a) that the media be legally required to publish specific technical information regarding public opinion surveys;

(b) that there be a blackout on publication of public opinion surveys by all print and electronic media during the final three days of an election campaign, in order to permit candidates adequate time to respond to controversial or misleading public opinion surveys before election day; and

(c) that a polling commission be established to ensure publication of complete methodological information, confidentiality of data, and public access to high quality information.

(Lachapelle’s affidavit, at para. 11.)

13 In its own Final Report, at p. 457, the Lortie Commission made this striking statement:

Although the industry in general has become highly professional since public polling was introduced in Canada in 1941, the incidence of technically deficient and poorly reported polls is still substantial. In recent elections, there have been instances of misleading polls, some because of technical errors and others because of partisan misrepresentation. There have even been allegations of fraudulent polls, where the data were said to have been fabricated to counter a poll showing the opposition in the lead. Such “bogus” polls and the more common misrepresented poll have been released to the media in many democracies. (Cantril 1991, 67; Worcester 1991, 199; Hoy 1989, 189-202) It is the willingness of the media to report such polls that makes them significant and troublesome. [Emphasis added.]

Regarding opinion poll regulation, the Lortie Commission recommended that there be, at the end of the election campaign, a 48-hour blackout period on publication of public opinion poll results. It also suggested technical information be provided as recommended by Lachapelle, but not the establishment of a polling commission.

14 An all-party Special Committee of the House of Commons on Electoral Reform considered the Lortie Commission’s Final Report. In its third report to the House of Commons, the Special Committee recommended a 72-hour prohibition period on publication of opinion survey results prior to the closing of the polls, but made no recommendation as to methodology information. The Special Committee’s recommendation of a 72-hour publication ban was incorporated into Bill C‑114, An Act to amend the Canada Elections Act. Bill C‑114 was adopted, without division, in the House of Commons on May 6, 1993: S.C. 1993, c. 19.

15 The Senate Standing Committee on Legal and Constitutional Affairs examined the legislation. Professor Peter Aucoin, Director of Research for the Lortie Commission, was a witness before the Senate Committee. He stressed the importance of polls during elections and submitted two arguments to support the proposed 72-hour ban:

While I support this particular proposal [the 72-hour blackout period], people must understand its purpose. The purpose is not, as has been said even in the Senate, to give voters a rest or breather from a flood of public opinion polls published just prior to election day. In fact, voters find that polls are useful information to them. They have a right to those polls. In terms of the electoral contest as we know it, the more polls the better.

The issue here is twofold. Public opinion polls purport to be scientific and are reported as such by the press. There is a question of the accuracy of public opinion polls in reading public opinion. There is clearly some need for regulation, given that particular assumption.

[P]arties and candidates need access to the polls and this requires some time. In particular, it means they have to be able to respond to the poll. They cannot do that if the poll is published either on election day or the day before an election. Therefore, the ban of 72 hours is to ensure that polls cannot be published after a point in time where candidates and parties cannot respond. It has nothing to do with giving voters a breather. [Emphasis added.]

(Senate Standing Committee on Legal and Constitutional Affairs, Proceedings, Issue No. 41, May 6, 1993, at pp. 41:14-41:15.)

Professor Aucoin also added the following as he responded to a comment from Senator Jean-Claude Rivest:

Senator Rivest: [translation] . . . I do not see how you can say that a poll is an important event but it must absolutely be checked in order to ensure that there are no bogus polls, no “hamburger polls”, as you explained. In this regard, in the last 72 hours there is a host of other election information and other statements that are not banned during this period and that cannot be verified before voting day.

Mr. Aucoin: [Text] I think the issue here is that the polls themselves in our culture and in our practice have a claim to scientific validity, notwithstanding the fact that they do have their limitations. In that context, to the degree that one makes a difference between published polls and just any other comments made by those participating in the election campaign, candidates and parties are at a disadvantage if there is an authenticity associated with the very concept of poll. Therefore, it has that character where one must be able to verify whether the poll exists and whether it is a credible poll. It is for those practical reasons that you would limit them.

If all polls were only done by certain kinds of organizations using certain kinds of standards, there would not be that problem. [Emphasis added.]

(Senate Standing Committee on Legal and Constitutional Affairs, supra, at pp. 41:17-41:18.)

16 The following incident serves to illustrate the usefulness of a ban on the publication of opinion poll results in the last 72 hours of an election campaign. In the midst of the 1993 federal election, the Globe and Mail made an error in its publication of the results of an Angus Reid poll (Globe and Mail, October 8, 1993). The Globe and Mail does not have a Sunday edition. If the erroneous article had been published on the last day of the campaign or on the Saturday prior to voting day, it would have been practically impossible for the pollster to convey the correct information, and for the newspaper to correct its report, before polling day. There would not have been an adequate opportunity to criticize and correct the inaccurate information before election day. The error was certainly not intentional, still it could have had far-reaching consequences, had it occurred in the last hours of an election campaign (Lachapelle’s affidavit, at paras. 30-34).

17 The appellants themselves acknowledge that opinion polls on election issues influence voters’ decisions: “[i]f voters paid no attention to polls there would be no point in suppressing them” (Appellants’ Factum, at para. 35). The very fact that voters pay attention to opinion surveys and rely on them as objective, non-partisan information makes it important that the information they convey not be misleading or inaccurate. The impugned provision responds to this need by requiring that election polls be published in sufficient time to allow for timely scrutiny and criticism.

18 I agree with my colleague that there are various activities subject to restriction during a federal election, that it is inappropriate to pronounce generally as to the scope or constitutional validity of these provisions, and that each type of restriction must be considered in its own context since each restriction is adopted to address a particular set of circumstances. However, these restrictions, including s. 322.1, stem from the same concerns. They are strong evidence that elections constitute, in our society, a unique event which calls for special treatment in order to promote voter autonomy and rational choice. Modern Canadian electoral law has sought to curb the excesses, enhance the democratic process and enable the voter to make a rational choice. Implicit to its regulations are the notions of integrity and fairness. As was aptly stated by the Ontario Court of Appeal in Re C.F.R.B. and Attorney-General for Canada, 1973 788 (ON CA), [1973] 3 O.R. 819, at p. 826:

Central to the whole democratic process is the election of legislative bodies by the vote of electors who by their ballots have expressed their choice, through a process, evolved over the years, designed to remove or at least reduce the possibility of the electors’ choice being unduly influenced by pressures put upon them.

Because of their claim to scientific validity, unscrutinized polls may have an influence that they do not actually deserve and may distort the electoral process.

II. Analysis

A. The Right to Vote (Section 3 of the Charter)

19 I agree with my colleague that a restriction on information would constitute an infringement of the right to vote under s. 3 of the Charter only if it undermines the guarantee of effective representation (Reference re Provincial Electoral Boundaries (Sask.), 1991 61 (SCC), [1991] 2 S.C.R. 158, at p. 183, McLachlin J.). In the instant case, I believe that the short blackout period has no such effect. On the contrary, such a period assists effective representation. A strategic voter cannot cast a significant vote if the information required to exercise that vote cannot be discussed and scrutinized in order to assess its real value. It follows that poll results which cannot be assessed in a timely manner may actually deprive voters of the effective exercise of their franchise. Therefore, there is no infringement of the right to vote under s. 3 of the Charter.

B. Freedom of Expression (Section 2(b) of the Charter)

20 It is not in dispute, and I agree with my colleague, that s. 322.1 infringes freedom of expression within the meaning of s. 2(b) of the Charter (R. v. Keegstra, 1990 24 (SCC), [1990] 3 S.C.R. 697). The issue then is whether this infringement constitutes, under s. 1 of the Charter, a reasonable limit demonstrably justified in a free and democratic society. I am in agreement with my colleague that this section serves a pressing and substantial objective, but must respectfully disagree with his conclusion that the section fails to minimally impair freedom of expression or is disproportionate in its effect. His conclusion rests on reasoning belied by the evidence and denies Parliament a choice of reasonable alternatives, holding it to a standard of perfection of uncertain reach. In the result, his reasons lead to the lifting of all restrictions as to the timing of the publication of opinion polls.

Justification Under Section 1 of the Charter

1. Preliminary Note: Evidence and Standard of Proof

21 In the course of the s. 1 analysis, the standard of proof to be used is proof on a balance of probabilities (R. v. Oakes, 1986 46 (SCC), [1986] 1 S.C.R. 103, at p. 137). As was stated in Libman v. Quebec (Attorney General), 1997 326 (SCC), [1997] 3 S.C.R. 569, at para. 39, scientific evidence is not required to meet the standard. Also McLachlin J., in writing for the majority in RJR-MacDonald Inc. v. Canada (Attorney General), 1995 64 (SCC), [1995] 3 S.C.R. 199, at para. 137, expressed the unanimous view of the Court:

Discharge of the civil standard does not require scientific demonstration; the balance of probabilities may be established by the application of common sense to what is known, even though what is known may be deficient from a scientific point of view. . . . [Emphasis added.]

To paraphrase the Court’s findings in Libman, supra, at para. 39, election campaigns, just as referendum campaigns, fall within the realm of social science, which does not lend itself to precise evidence.

2. Legislative Objective

22 In the material before the Court, reference is made to three suggested legislative objectives: first, to reduce the “undue influence” of all polls in general; second, to provide for a rest period so as to allow electors time to reflect before polling day; and third, to prevent the potentially distorting effect of public opinion survey results that are released late in an election campaign leaving insufficient time to assess their validity. This latter objective is the only one that the respondent advanced before this Court. Although my analysis will focus on this particular objective, the first two objectives may not be without merit. However, they need not be considered here. As I mentioned earlier, opinion polls have reshaped Canadian elections. It is now a fact that election campaigns take on an aura of “horse races”, and that discussion of issues that concern Canadians tends to be preempted. Voters are of course completely free to choose the information upon which they want to make their decision. For instance, strategic voters, who may want to vote for their second-choice candidate in order to avoid the election of a leading candidate, rely on polls. Thus, while suppressing polls simply because they may be used by voters is not permissible, regulating polls may be.

(a) Criteria

23 The Court must first assess the objective of the infringing legislative measure, as distinguished from the means chosen to implement it. The question is whether the concern which prompted the enactment of the impugned legislation is pressing and substantial and whether the purpose of the legislation is one of sufficient importance (Irwin Toy Ltd. v. Quebec (Attorney General), 1989 87 (SCC), [1989] 1 S.C.R. 927, at p. 987, Dickson C.J. and Lamer J. (as he then was) and Wilson J.). The distinction between “objective” and “means” is important since at this stage, the Court must ensure that the said objective is consistent with the principles integral to a free and democratic society, pressing and substantial, and directed to the realisation of collective goals of fundamental importance (Oakes, supra, at p. 138).

24 In Oakes, supra, at p. 136, Dickson C.J. stated that in determining whether Charter rights and freedoms should be limited,

[t]he Court must be guided by the values and principles essential to a free and democratic society which I believe embody, to name but a few, respect for the inherent dignity of the human person, commitment to social justice and equality, accommodation of a wide variety of beliefs, respect for cultural and group identity, and faith in social and political institutions which enhance the participation of individuals and groups in society. The underlying values and principles of a free and democratic society are the genesis of the rights and freedoms guaranteed by the Charter and the ultimate standard against which a limit on a right or freedom must be shown, despite its effect, to be reasonable and demonstrably justified. [Emphasis added.]

As La Forest J. put it in Ross v. New Brunswick School District No. 15, 1996 237 (SCC), [1996] 1 S.C.R. 825, at para. 77, ultimately, any attempt to determine whether the impugned limit is a justifiable infringement of the freedom of expression must involve a weighing of the essential principles of a free and democratic society. The enhancement of the electoral process, through the provision of quality, timely information, serves one of these fundamental principles, i.e., good participation and faith of individuals in the most important political institution of all ^__ the electoral process.

(b) Purposes of Freedom of Expression: Access to Information

25 As to the purposes underlying freedom of expression, La Forest J. in Ross, at para. 59, and McLachlin J. in R. v. Zundel, 1992 75 (SCC), [1992] 2 S.C.R. 731, at p. 752, noted that s. 2(b) of the Charter aims at permitting free expression “to the end of promoting truth, political or social participation, and self-fulfilment”. Following these purposes, freedom of expression should not be considered as an end per se. The promotion of an informed vote over a misinformed vote meets the three purposes and truly serves the core values of the freedom of expression in a free and democratic society: by allowing timely discussion of all published poll results, s. 322.1 aims at fostering truth; by keeping open the possibility of timely debate as to the validity of poll results, it promotes active political and social participation, rather than condone passiveness as to poll results; by allowing for full scrutiny of the information carried by poll results late in the election campaign, it promotes voters’ self-fulfilment by ensuring that the intention voters really want to convey in casting their vote is actually expressed.

26 The quest for better information gives more meaning to voter participation in the electoral process. The very fact that some voters base their decision on opinion survey polls may justify the means taken to promote voters’ right to good information. This is consistent with the findings of this Court that one of the objectives underlying freedom of expression is the ability of voters to make informed choices (Libman, supra, at para. 54; Ford v. Quebec (Attorney General), 1988 19 (SCC), [1988] 2 S.C.R. 712, at p. 767).

27 As a matter of fact, in 1979, the committee on polling of the Regroupement québécois des sciences sociales suggested ways to improve the use of polls by journalists and to encourage increased accessibility to poll results broadcast during referendum campaigns. Quebec academics were concerned that with so many polls being conducted the public would not be able to distinguish between “good” and “poor” polls in the heat of the moment and that certain minimum standards of polling would lose out to partisan considerations. One of the committee’s recommendations was that publication and broadcast of opinion polls during the week prior to the election be prohibited. Public discussion on published surveys, however, would still be allowed (Lachapelle Study, supra, at pp. 40-41). The foregoing recommendations were analysed by the Commission des droits de la personne du Québec in the light of the Quebec Charter of Human Rights and Freedoms, R.S.Q., c. C-12. As Lachapelle noted at pp. 45-46:

According to the human rights commission of Quebec, three basic corollaries affect the public’s right to be fully informed, and these must be respected in the context of an election or referendum campaign:

1. the availability of accessible, unrestricted information;

2. access to plentiful, diversified information; and

3. access to accurate, high-quality information, which implicitly guarantees the freedom to criticize that information.

As to the appropriateness of a blackout period, Lachapelle mentioned that the Commission des droits de la personne du Québec expressed reservations and urged public debate on the issue, but decided against taking a position. According to Lachapelle, “[t]he reasons for the commission’s indecision were that although such a prohibition would interfere with freedom of expression, it might also encourage public debate and afford voters a better opportunity to consider the real issues at stake. The commission also emphasized that in a system with no restrictions, it would be impossible to correct any erroneous information published or broadcast during the final hours of a campaign. In short, this proposal would restrict the right to information, as defined in the commission’s first two principles [corollaries], in order to promote the third.” (Lachapelle Study, at p. 46.)

28 The position of the appellants and the intervener Canadian Civil Liberties Association regarding free expression in democracy is couched on the rationale that truth emerges through vigorous debate and more publication of polls. My colleague adopts this view, at para. 108:

But an opinion poll does not appear in a vacuum. Rather, it is published chronologically after a series of other polls which have been measuring public opinion throughout the election. In all likelihood, other polls conducted by other polling organizations will appear in other media outlets during the three days prior to election day. . . . The more polls which appear during this period, the less likely that voters will base their decisions on the inaccurate poll.

This philosophical underpinning was expressed by McIntyre J., speaking for the majority, in RWDSU v. Dolphin Delivery Ltd., 1986 5 (SCC), [1986] 2 S.C.R. 573, where he wrote at p. 583:

The importance of freedom of expression has been recognized since early times: see John Milton, Areopagitica; A Speech for the Liberty of Unlicenc’d Printing, to the Parliament of England (1644), and as well John Stuart Mill, “On Liberty” in On Liberty and considerations on Representative Government (Oxford 1946), at p. 14:

If all mankind minus one were of one opinion, and only one person were of the contrary opinion, mankind would be no more justified in silencing that one person, than he, if he had the power, would be justified in silencing mankind.

And, after stating that “All silencing of discussion is an assumption of infallibility”, he said, at p. 16:

Yet it is as evident in itself, as any amount of argument can make it, that ages are no more infallible than individuals; every age having held many opinions which subsequent ages have deemed not only false but absurd; and it is as certain that many opinions now general will be rejected by future ages, as it is that many, once general, are rejected by the present.

Nothing in the vast literature on this subject reduces the importance of Mill’s words. The principle of freedom of speech and expression has been firmly accepted as a necessary feature of modern democracy.

According to this rationale, if there are distorting effects, these should be dealt with through corrective response of civil society (e.g., the media, political parties, private individuals, etc.), not through coercive action by the state. There is a serious problem with relying unreservedly on this rationale when it comes to elections. To say that truth most reliably emerges by means of correction through more polls is to assume an ongoing debate. In elections, the debate ends with the vote. A multiplicity of potentially inaccurate polls, none of which are in time to permit debate, fosters confusion and offers little protection to the public. Errors and misinformation may be corrected after the election, but the value of the correction is lost. Elections suggest that a special remedy may be in order: namely, a requirement that information be timely so as to avoid the harm occurring in the first place.

(c) Purpose of Section 322.1

29 The purpose of s. 322.1 is to improve information to the public during election campaigns. Its primary objective is positive rather than negative. Section 322.1 does not purport to suppress an evil per se, such as obscenity in R. v. Butler, 1992 124 (SCC), [1992] 1 S.C.R. 452, or hate propaganda in Keegstra, supra. It aims at balancing and enhancing Charter rights, namely the informed exercise of the right to vote and a fundamental purpose of freedom of expression, i.e., informed participation in the electoral process. If one wants to identify a harm in the instant case, it would be a lack of enhancement of the electoral process through timely information. Section 322.1 aims at improving the search for truth, by providing for the timeliness of the publication of poll results, so as to allow discussion, not simply proscribing polls. It in no way dictates or deals with the content of expression.

30 The Charter should not become an impediment to social and democratic progress. It should not be made to serve substantial commercial interests in publishing opinion poll results, by defeating a reasonable attempt by Parliament to allay potential distortion of voter choice.

31 Parliament adopted s. 322.1 without expressed opposition, following lengthy, extensive, in-depth studies and consideration over several decades, in response to a long-standing concern. Being themselves the very objects of elections, members of Parliament were in the best position to assess the effects of polls in electoral campaigns and their impact on individual voters. In the proceedings, it has not been suggested that members of Parliament had any interest other than to foster the integrity of the electoral process by avoiding voters being misled on facts through denial of an opportunity for scrutiny and discussion.

(d) Evidence

32 Social science studies composed much of the evidence submitted before the motions judge. In that respect, Somers J. rightly stated, at p. 142:

[I]t would be a mistake to look only at social science evidence in assessing whether the objective is pressing and substantial. . . . Of course, social science data are ready evidence of the importance of different problems, but it is not the only evidence courts should consider, nor should it be given decisive weight or considered requisite to government action. Thus, even if social science evidence is found to be inconclusive, it is still possible to ground a pressing objective on evidence of extended public debate of an issue or by looking at the actions of other democracies. On this application, it is clear from the number of times a concern about polling was raised by the public and in reports stemming from public consultation, that many considered this to be a pressing issue. . . . In a democracy, it is to be expected that representatives will react to such a debate, either through private members bills or government legislation; it might be an institutional failure if they did not. [Emphasis added.]

33 As outlined earlier in these reasons, Somers J. had ample material to support his position that, taken as a whole, the evidence “serves to buttress the respondent’s contention that the unregulated distribution of poll results is not without its potential problems” (p. 121). It must be pointed out that the Court of Appeal unanimously accepted Somers J.’s observations and findings of facts regarding his s. 1 analysis (see (1996), 1996 890 (ON CA), 30 O.R. (3d) 350, at pp. 359-60). The evidence shows that there exists a long-standing concern about the publication of opinion survey results during election campaigns in Canada. In the last 30 years at least, several reports have studied and discussed this issue in depth.

34 In the instant case, at p. 143, Somers J. rightly found that

it is reasonable to presume that polls can harm the electoral process. I rely here on the following evidence:

(1) the sheer prevalence of polling results (22 national polls reported during the 1988 election, mentioned in some 30 per cent of television coverage);

(2) awareness of poll results amongst 70 to 80 per cent of the public;

(3) methodological information necessary of critical evaluation is very frequently absent from reporting;

(4) the lack of a corrective response time when polls are published close to election day.

In reaching these conclusions, Somers J. makes specific reference to several findings of the Lortie Commission. Referring to its finding, at p. 457 of its Final Report, that “recent research provides strong support for the proposition that published opinion polls can significantly influence campaigns and voters”, he commented (at p. 118) that “[t]he Commission based this conclusion on the fact that ‘[o]ur research demonstrates that polls did have measurable effects on the conduct of election campaigns and the choice voters make’ (p. 456).”

35 As he stated, at p. 119:

. . . the evidence presented dealt not only with the effect of polls on voters, but also with problems associated with the undue influence, late publication and accuracy of polls. The evidence pointed to a general worry about the “undue influence” of opinion polls on the election process. This influence is said to flow from two related factors. The first is that the polls are presented as scientific and authoritative. The second is that despite this presentation, the public is often not given the information needed to gauge a poll’s true accuracy. [Emphasis added.]

This is further explained in the Lortie Commission Final Report (p. 455 quoted at p. 119 of the judgment):

Because they are presented as “scientific”, published opinion polls raise issues of public confidence in the integrity of the electoral process. Notwithstanding their claims to scientific validity and accuracy in representing the views of all potential voters, opinion polls are susceptible to many forms of error and misrepresentation. The apparent precision of the data they report fails to reflect the fact that they are estimates of the distribution of opinion at a given time. Yet their apparent authority gives them considerable influence over the conduct of campaigns and the choices made by voters. [Emphasis added.]

36 Somers J. also found, at p. 142, that the international review of legislation regarding publication of opinion poll results during elections was “strong evidence of a pressing objective”. At least six European Union countries have some form of polling blackout, ranging in duration from five days in Spain to the entire length of the election in Portugal. French regulations prohibit publication during the last week before a round of voting for presidential, legislative, or European elections. In Belgium, the prohibition lasts 30 days. Greece, Brazil, South Africa, South Korea, Japan, and Australia enacted legislation to regulate or ban the publication of poll results. The Lachapelle Study, supra, concluded its international survey, at p. 68, by asserting that:

A review of the legislation of various countries reveals that governments tend toward outright banning of polls near the end of election campaigns rather than requiring the publication of a specifications sheet. Only France has chosen both options. However, every law is imperfect and can only partially regulate as multifaceted a practice as polling. [Emphasis added.]

The fact that so many democratic, industrialized nations have passed legislation in this particular area is further evidence of the importance of the problem. Extensive polling is a relatively recent phenomenon. That some countries have not responded does not detract from the fact that many have. To make the response of others to political or social concerns a pre-requisite to legislation would transform the Charter into a sword against social progress and stultify government action.

37 In their factum, the appellants refer to a decision of the Alberta Court of Appeal, Canada (Attorney General) v. Somerville, 1996 ABCA 217, [1996] 8 W.W.R. 199, to support the view that Parliament’s objective as to the impugned legislation is not permissible under the Charter. In Somerville, the Court of Appeal held that certain provisions of the Canada Elections Act which, among other things, limited third party advertising expenses during a federal election campaign to a maximum of $1,000 per individual, were contrary to freedom of expression under the Charter. Conrad J.A., Harradence J.A. concurring, stated, at pp. 228 and 231, that:

The Attorney General argues that unrestricted third party advertising could distort the political process.

This legislation bans input. This is a case where the objective of the legislation is not trying to balance expenditures of outside groups, the press and parties. Rather, one is led to conclude that the very aim or purpose of this legislation is to ensure that third parties cannot be heard in any effective way and that political parties are entitled to preferential protection. Its objective strikes at the core of these fundamental rights and freedoms, and is arguably legislation which has as its very purpose the restriction of these rights and freedoms, which can never be justified. [Emphasis added by the appellants.]

In Libman, supra, at para. 56, the Somerville decision was unanimously criticized inasmuch as this Court declared “that the objective of Quebec’s referendum legislation [of promoting fairness in a democratic process through a certain equality of resources] is highly laudable, as is that of the Canada Elections Act”. In Libman, the system set up by the referendum legislation restricted independent spending, thus freedom of expression, in order “to preserve a balance in the promotion of the options and favour an informed and truly free exercise of the right to vote” (para. 54 (emphasis added)). The instant case aims at this very objective.

38 Considering that the legislative objective of s. 322.1 of the Canada Elections Act is consistent with and indeed enhances the objectives underlying freedom of expression, among them the ability of voters to make informed choices and the promotion of political and social participation, that Parliament’s concern is clearly pressing and substantial, in that it is directed to the realisation of the important collective goal of safeguarding the integrity of the electoral process (Harvey v. New Brunswick (Attorney General), 1996 163 (SCC), [1996] 2 S.C.R. 876, at para. 38), I conclude that the objective of preventing the potentially distorting effect of public opinion survey results that are released late in an election campaign when there is no longer a sufficient opportunity to respond is a sufficiently important objective which meets the first step of the analysis under s. 1 of the Charter.

3. Proportionality Test

(a) Rational Connection

39 Parliament has chosen to ban the publication, dissemination and broadcast of poll results from midnight on Friday before polling day through the end of polling day. “The essence of rational connection is a causal relationship between the objective of the law and the measures enacted by the law. This is often a difficult matter to establish by evidence, and the Supreme Court of Canada has not always insisted on direct proof of the causal relationship.” (P. W. Hogg, Constitutional Law of Canada (loose-leaf ed.), vol. 2, at p. 35-29.) In RJR-MacDonald, supra, the Court unanimously agreed that a causal relationship between advertising and tobacco product consumption could be based either on common sense, reason, or logic (La Forest J., at para. 86, McLachlin J., at paras. 156-58, and Iacobucci J., at para. 184), even though the evidence may be admittedly inconclusive. In Butler, supra, Sopinka J. found, at p. 502, that it was “reasonable to presume” that there is a causal relationship between obscenity and harm to society. Similarly, in Ross, supra, La Forest J., writing for the Court, held, at para. 101, that it was “reasonable to anticipate” that there is a causal link between anti-Semitic activity by school teachers outside school and discriminatory attitudes within school.

40 In the instant case, the rational connection is self-evident. Opinion polls significantly influence voter choice and electoral campaigns. It follows that the publication of inaccurate, though authoritative, opinion survey results that go uncorrected may well lead to voters making misinformed decisions. Logically, there is a reasoned apprehension that voters will be deprived of the full exercise of their franchise. Its importance is measured by the significant influence of polls on voters and the prevalence of misleading polls. Ensuring that polls that cannot be adequately, publicly and independently evaluated as to their correctness because of insufficient time are not published clearly addresses this problem. Voters are free to cast their ballot as they see fit; however, the democratic process cares about each voter and should not tolerate the fact that, in the pooling booth, some voters would express themselves on the basis of misleading, or potentially misleading, information that is de facto immunized from scrutiny and criticism.

(b) Minimal Impairment

41 In my view, there is no doubt that s. 322.1 constitutes a genuine mediation between the rights of voters to receive information in a timely fashion, and the right of pollsters and publishers freely to provide the information they want. Not only does the legislation protect the rights of voters but it does so by serving one of the very purposes of freedom of expression, that is informing while allowing for political debate and discussion. As mentioned earlier, the provision also strikes a balance between two basic aspects of voters’ right to information, that are, on the one hand, the availability of accessible, unrestricted, plentiful, diversified information and, on the other, the timely availability of factual information that may be misleading so as to allow for scrutiny and criticism.

42 In matching means to ends and asking whether rights are impaired as little as possible, a legislature mediating between the claims of competing groups will be forced to strike a balance without the benefit of absolute certainty concerning how that balance is best struck (Irwin Toy, supra, at p. 993). As the majority also held in Irwin Toy, at p. 999: “This Court will not, in the name of minimal impairment, take a restrictive approach to social science evidence and require legislatures to choose the least ambitious means to protect vulnerable groups.” I share the respondent’s view that this Court should not second-guess the wisdom of Parliament in its endeavour to draw the line between competing credible evidence, once it has been established, on the civil standard of proof, that Parliament’s objective was pressing and substantial:

When striking a balance between the claims of competing groups, the choice of means, like the choice of ends, frequently will require an assessment of conflicting scientific evidence and differing justified demands on scarce resources. Democratic institutions are meant to let us all share in the responsibility for these difficult choices. Thus, as courts review the results of legislature’s deliberations, particularly with respect to the protection of vulnerable groups, they must be mindful of the legislature’s representative function.

(Irwin Toy, at p. 993.)

43 At this stage, the question is whether there is a reasonable basis, on the evidence tendered, for concluding that a blackout on all opinion polls during the last weekend of an election campaign and during election day impaired freedom of expression as little as possible given the government’s pressing and substantial objective (Irwin Toy, at p. 994). Parliament is not bound to find the least intrusive nor the best means. This would be too high a standard for our elected representatives to meet. In RJR-MacDonald, supra, at para. 160, McLachlin J. stated:

[T]he government must show that the measures at issue impair the right of free expression as little as reasonably possible in order to achieve the legislative objective. The impairment must be “minimal”, that is, the law must be carefully tailored so that rights are impaired no more than necessary. The tailoring process seldom admits of perfection and the courts must accord some leeway to the legislator. If the law falls within a range of reasonable alternatives, the courts will not find it overbroad merely because they can conceive of an alternative which might better tailor objective to infringement. . . . [Emphasis added.]

(See also Harvey v. New Brunswick (Attorney General), supra, at para. 47, per La Forest J.)

In Keegstra, supra, Dickson C.J. also noted at pp. 784-85:

In assessing the proportionality of a legislative enactment to a valid governmental objective, however, s. 1 should not operate in every instance so as to force the government to rely upon only the mode of intervention least intrusive of a Charter right or freedom. It may be that a number of courses of action are available in the furtherance of a pressing and substantial objective, each imposing a varying degree of restriction upon a right or freedom. In such circumstances, the government may legitimately employ a more restrictive measure, either alone or as part of a larger programme of action, if that measure is not redundant, furthering the objective in ways that alternative responses could not, and is in all other respects proportionate to a valid s. 1 aim. [Emphasis added.]

44 One can conceive of alternatives to the impugned measure. In RJR-MacDonald, supra, at para. 160, McLachlin J. stated: “if the government fails to explain why a significantly less intrusive and equally effective measure was not chosen, the law may fail” (emphasis added). I conclude that these alternative measures cannot pretend to equally serve the objective of preventing the potentially distorting effect of public opinion survey results that are released late in an election campaign when there is no longer a sufficient opportunity to respond, and be as effective when compared to the measure at issue.

45 One of the suggested alternatives is mandatory publication of methodological information. Since poll results would be allowed to be published up until the last minute of the election campaign, it supposes that voters can, looking at that methodological information, decide for themselves if the poll was properly conducted and ascertain whether the results are reliable. Absent sufficient time for public discussion of the poll results, it is unlikely that this alternative would be of real assistance as voters may lack the requisite knowledge to properly assess the results. The publication of methodological information is useful when analysts and political parties have sufficient time to evaluate its validity.

46 The second proposed solution is to create a punishment for publishing false poll results. Once again, the proposal suffers from serious shortcomings. Firstly, the prohibition is unlikely to prevent voters from being misled. Despite successful prosecution, the damage to voters would already have been done. Secondly, even if one were to assume that such a prohibition, coupled with severe sanctions, would serve as a deterrent, it would only aim at intentional deception. Consequently erroneous polls whose results will not have been considered due to lack of time will still mislead voters and remain unpunished.

47 In summary, there is simply no equally effective alternative to the current short-term blackout for achieving the legislative objective.

48 What is then the scope of s. 322.1? It is claimed that s. 322.1 is overbroad in that it prohibits polls of all kinds regardless of their scientific nature or quality. I agree with the Court of Appeal’s finding that the provision prohibits all opinion surveys, regardless of how informal they may be, “hamburger polls” included. The expression “hamburger polls” describes what is sometimes used as a marketing device during elections where goods are matched to political candidates or parties. The sales of each item are then claimed to reflect public sentiment. As amusing as it may sound, the expression “hamburger polls” tends to suggest a clear cut line between reliable poll results and misleading poll results. There is no such clear demarcation. To limit the ban to polls that are acceptable to the scientific community leaves out polls that, some voters may wrongly be led to believe, are representative or based on scientific methodology, and complicates enforcement of the provision. On the other hand, the information conveyed by “hamburger polls” is of such questionable nature that any infringement of Charter rights is minimal, at best.

49 I add that, considering the main legislative objective, I share my colleague’s view that the impugned legislation does not apply to the discussion of previously released poll results. After stating two purposes for s. 322.1, namely to provide voters with a rest period and to guard them against an inaccurate poll which occurs late in the campaign, my colleague writes, at para. 99:

Accepting these as the two objectives of s. 322.1, it is my view that any ambiguity in the words of the section should be interpreted in accordance with those purposes, rather than to frustrate them. I conclude, therefore, that Somers J. was correct in interpreting the prohibition in s. 322.1 as applying only to “new” poll results, i.e. results that are undisclosed as of midnight on the Friday before election day. Far from preventing old poll results from being mentioned, the very raison d’être of the section is that those old poll results should be aired and discussed in the media so that their accuracy can be fully determined in public debate.

The legislation prohibits broadcast, publication and dissemination. Ejusdem generis, these expressions refer only to the initial release of poll results. In comparison, French and Belgian regulations more clearly ban the publication, broadcast and commentary of the surveys.

50 The impugned limitation minimizes the risks of publication and dissemination of misleading poll results on or just before the crucial moment of the polling day, by allowing just enough time (72 hours approximately), from Friday prior to election day to the closing of polling stations on election day, to collect any undisclosed methodological information, assess poll results, discuss the assessment, criticize the analyses, and disseminate the results of the discussions throughout the electorate. This period is very shortlived, especially for the analysis of a cross-Canada opinion poll and the airing of the analysis, having regard to the time allotted for voting and the restriction on broadcasting and advertising on polling day and the day preceding.

51 The Lachapelle Study suggested 72 hours, while the Lortie Commission recommended 48. Professor Lachapelle submitted, at p. 157 of his study, that:

The 72-hour blackout suggested here would put print and broadcast journalists on the same footing and would respond to a long-standing grievance among broadcast journalists that they face more restrictions than their print colleagues. The legislator could envisage eliminating the 48-hour rule entirely, but such a route might jeopardize a relative consensus on the need to limit all partisan information at the end of a campaign. A supplementary 24 hours would enable citizens to exercise their moral right to reply, which would encourage public discussion, especially when it is a matter of deciding who will govern us for the next few years. Broadcasters and citizens should, therefore, draw dividends from this measure.

The 48-hour rule mentioned by Professor Lachapelle refers to s. 213(1) of the Canada Elections Act which provides that:

213. (1) Any person is guilty of an offence who, for the purpose of promoting or opposing a particular registered party or the election of a particular candidate, directly or indirectly,

(a) . . . the one day immediately preceding polling day or on polling day, advertises on the facilities of any broadcasting undertaking; or

(b) procures for publication or acquiesces in the publication, . . . on the one day preceding polling day or on polling day of an advertisement in a periodical publication.

In Irwin Toy, supra, at issue was the determination of the upper age limit for the protection of children from advertising. This Court held that the legislature was not obliged to confine itself solely to protecting the most clearly vulnerable group. It was only required to exercise reasonable judgment in specifying the vulnerable group. The Court quoted Dickson C.J. in R. v. Edwards Books and Art Ltd., 1986 12 (SCC), [1986] 2 S.C.R. 713, at pp. 781-82:

I might add that I do not believe there is any magic in the number seven as distinct from, say, five, ten, or fifteen employees as the cut‑off point for eligibility for the exemption. In balancing the interests of retail employees to a holiday in common with their family and friends against the s. 2(a) interests of those affected the Legislature engaged in the process envisaged by s. 1 of the Charter. A “reasonable limit” is one which, having regard to the principles enunciated in Oakes, it was reasonable for the legislature to impose. The courts are not called upon to substitute judicial opinions for legislative ones as to the place at which to draw a precise line. [Emphasis added.]

Based upon the current legislation, the reports and the studies available, Parliament reasonably determined, as it was entitled to do, that a 72-hour period was necessary to allow meaningful scrutiny of poll results during an election campaign.

52 Furthermore, the limitation period only affects one mode of expression: opinion poll results respecting how electors will vote at an election or respecting an election issue that would permit the identification of a political party or candidate. This mode of expression is not a primary source of information concerning relevant political facts. It mainly constitutes information as to the effect of relevant political information on potential voters. Candidates and political parties, through the media, are still allowed to disseminate their program across the electorate and to carry on their campaign, within the other limits prescribed by the Canada Elections Act, which are not currently at issue.

53 Many elements of information can factor into influencing voters: statistics, financial information, speeches, news, etc. When Parliament identifies one matter of concern, it has no absolute duty to identify and regulate each and every factor. It is entitled to determine the urgency of addressing a particular problem and the appropriate means of doing so. As La Forest J. put it for the majority in McKinney v. University of Guelph, 1990 60 (SCC), [1990] 3 S.C.R. 229, at pp. 317-18:

In looking at this type of issue, it is important to remember that a Legislature should not be obliged to deal with all aspects of a problem at once. It must surely be permitted to take incremental measures. . . . This Court has had occasion to advert to possibilities of this kind. In R. v. Edwards Books and Art Ltd. [1986 12 (SCC), [1986] 2 S.C.R. 713], Dickson C.J., there dealing with the regulation of business and industry, had this to say, at p. 772:

I might add that in regulating industry or business it is open to the legislature to restrict its legislative reforms to sectors in which there appear to be particularly urgent concerns or to constituencies that seem especially needy. In this context, I agree with the opinion expressed by the United States Supreme Court in Williamson v. Lee Optical of Oklahoma, 348 U.S. 483 (1955), at p. 489:

Evils in the same field may be of different dimensions and proportions, requiring different remedies. Or so the legislature may think. . . . Or the reform may take one step at a time, addressing itself to the phase of the problem which seems most acute to the legislative mind. . . . The legislature may select one phase of one field and apply a remedy there, neglecting the others. [Emphasis added.]

Both the motions judge and the Court of Appeal held that inaccurate polls at the end of an election campaign constitute a reasonable concern. The respondent has not to show that this concern is more serious or is causing more harm to the electoral process or to individual voters than any other potentially misleading information. There is no such standard under the Charter. Holding so would have the effect of transforming the Charter into an impediment to social progress. As Somers J. found, studies and reports, as well as bills in the House of Commons, and legislation in other democratic countries support Parliament’s reasonable finding that the concern at bar was serious.

54 My colleague asserts, at paras. 112 and 113, that:

In this case, however, the social science evidence did not establish that the Canadian voter is a vulnerable group relative to pollsters and the media who publish polls. The presumption in this Court should be that the Canadian voter is a rational actor who can learn from experience and make independent judgments about the value of particular sources of electoral information. . . . However, no evidence has been presented before this Court that voters have suffered from any misapprehensions regarding the accuracy of any single poll. Indeed, the fact that polls conducted contemporaneously yield differing results, or that poll results can fluctuate dramatically over time, suggests that voters have experience with the shortcomings of some polls. . . . Voters are constantly exposed to opinion poll results throughout the election and a single inaccurate poll result is likely to be spotted and discounted appropriately.

. . . What I have said in the previous paragraph suggests, as a matter of logic, that there is reason to believe that, notwithstanding the scientific “aura” of polls, the Canadian voter is likely to be aware of a seriously inaccurate poll. Indeed, the more serious the inaccuracy, the more likely the awareness of the error. [Emphasis added.]

With respect, I must differ for two reasons. While voters may be credited with some knowledge of the reliability of poll results generally, surely it is legitimate for Parliament to provide them with an opportunity of distinguishing “poor” from “good” polls. As I mentioned earlier, it is concern as to the prevalence of the publication of “poor” polls which has prompted the legislation. An opportunity to analyse, discuss and criticize published poll results is required in all cases, be it by any individual voter or more likely by analysts and others, in sufficient time for a public airing of diverse opinions. In addition, while my colleague acknowledges the pressing and substantial legislative objective and its rational connection to the impugned measure, he apparently finds the evidence submitted by the government insufficiently specific and conclusive to justify the impairment. The effect of potentially inaccurate opinion poll results must be measured having regard to the effect of opinion poll results generally. Reasoning based on the belief that voters will show discernement in their reliance on polls and that only some voters are influenced does not belie nor answer the fact that the influence of polls both on the choice of voters and on the conduct of electoral campaigns is significant.

55 My colleague also asserts, at para. 117:

The Canadian voter is not a historically vulnerable or disadvantaged group. Nor, as has been explained above, is the autonomy or dignity of any single group under attack from, or even facing the contrary interests of, another potentially more powerful group. Nor can it be said that there is a shared understanding amongst Canadians that a single inaccurate poll will mislead Canadians to an extent which . . . is “undue”. I am, therefore, unable to accept that the harm which the government is seeking to prevent affects a large number of voters, or that such possible distortions are significant to the conduct of an election, without more specific and conclusive evidence to that effect. [Emphasis added.]

56 With respect, everyone is vulnerable to misinformation which cannot be verified. As I have pointed out, a multiplicity of potentially inaccurate polls offers little protection to the public. Our democracy, and its electoral process, finds its strength in the vote of each and every citizen. Each citizen, no matter how politically knowledgeable one may be, has his or her own reasons to vote for a particular candidate and the value of any of these reasons should not be undermined by misinformation.

57 The findings of the Lortie Commission in its Final Report (at p. 457) bear repeating:

Although the industry in general has become highly professional since public polling was introduced in Canada in 1941, the incidence of technically deficient and poorly reported polls is still substantial. In recent elections, there have been instances of misleading polls, some because of technical errors and others because of partisan misrepresentation. There have even been allegations of fraudulent polls, where the data were said to have been fabricated to counter a poll showing the opposition in the lead. Such “bogus” polls and the more common misrepresented poll have been released to the media in many democracies. (Cantril 1991, 67; Worcester 1991, 199; Hoy 1989, 189-202) It is the willingness of the media to report such polls that makes them significant and troublesome. [Emphasis added.]

58 In other words, opinion poll results have a significant influence. They are subject to error, misrepresentation and open to manipulation. Elections, by their very nature, are a battleground for diverging interests, each striving for the favour of electors. Polls are one instrument in the armoury of the contestants. The intense conflicts of interest in an election are an inducement to manipulation. Parliament, acting on the conclusions of extensive studies and research after consideration over many years and in response to a broad public concern with the benefit of the background experience of its own members, has chosen to enact a modest measure of protection of voters against factual misinformation.

(c) Proportionate Effects

59 In Dagenais v. Canadian Broadcasting Corp., 1994 39 (SCC), [1994] 3 S.C.R. 835, at p. 889, Lamer C.J. rephrased the third part of the Oakes test of proportionality:

. . . there must be a proportionality between the deleterious effects of the measures which are responsible for limiting the rights or freedoms in question and the objective, and there must be a proportionality between the deleterious and the salutary effects of the measures. [Emphasis in original.]

60 The salutary effect of s. 322.1 is to promote the right of voters not to be misled in the exercise of their right to vote. At the end of an election campaign, opinion poll results have the potential to irreparably mislead voters, particularly strategic voters, because of the purportedly scientific, reliable nature of most opinion poll results. The deleterious effect of the measure is that s. 322.1 deprives some voters, who rely on polls to make their decision, of some late campaign opinion poll results. This deleterious effect is quite limited, when one considers the delay between conducting the poll and ultimately publishing its results. In fact, what is prohibited is the publication of results of polls conducted at best three days or more before polling day, as against polls conducted five or more days before polling day. And as we can see from Table 1.11 of the Lachapelle Study, supra, at pp. 113-15, a polling organization could take up to seven days to interview their respondents for a cross-Canada survey. This tends to detract from both the accuracy and the timeliness of the “snapshot”. As I pointed out earlier, a strategic voter cannot cast a significant vote if the information required to exercise that vote is not discussed and scrutinized in order to assess its real value. Poll results which cannot be assessed in a timely manner may actually deprive voters of the effective exercise of their franchise.

61 As to the effects of the measure on freedom of expression, on the one hand, s. 322.1 precludes the media from publishing polls on the last weekend of the election campaign and on polling day. This ban causes minimal impairment to freedom of expression because of its very short duration and because of the lack of satisfactory alternatives available to tailor the measure to the legislative objective. On the other hand, s. 322.1 has a positive impact on freedom of expression. It promotes debate and truth in political discussion since it gives voters the opportunity to be informed about the existence of misleading factual information. The salutary effects of the measure concerning both the right to vote and freedom of expression thus outweigh the deleterious effects caused by the impugned provision.

III. Conclusion

62 I therefore find that s. 322.1 of the Canada Elections Act does not infringe the right to vote as guaranteed under s. 3 of the Charter and that, while it restricts freedom of expression within the meaning of s. 2(b) of the Charter, it constitutes a reasonable limit demonstrably justified in a free and democratic society under s. 1 of the Charter and hence does not violate the Charter.

IV. Disposition

63 For the above reasons, I would answer the constitutional questions posed by Lamer C.J. as follows:

1. Does s. 322.1 of the Canada Elections Act, R.S.C., 1985, c. E-2, as amended, infringe s. 2(b) and/or s. 3 of the Canadian Charter of Rights and Freedoms?

Answer: Section 322.1 of the Canada Elections Act does not infringe s. 3 of the Charter but infringes s. 2(b) of the Charter.

2. If s. 322.1 of the Canada Elections Act infringes s. 2(b) and/or s. 3 of the Canadian Charter of Rights and Freedoms, is s. 322.1 a reasonable limit prescribed by law as can be demonstrably justified in a free and democratic society for the purposes of s. 1 of the Canadian Charter of Rights and Freedoms?

Answer: The infringement of s. 2(b) of the Charter by s. 322.1 is a reasonable limit prescribed by law as can be demonstrably justified in a free and democratic society for the purposes of s. 1 of the Charter.

64 I would therefore dismiss the appeal with costs throughout.

//Bastarache J.//

The judgment of Cory, McLachlin, Iacobucci, Major and Bastarache JJ. was delivered by

65 Bastarache J. -- This appeal concerns the constitutional validity of s. 322.1 of the Canada Elections Act, R.S.C., 1985, c. E-2, which prohibits the broadcasting, publication, or dissemination of opinion survey results in the final days of a federal election campaign. More specifically, the provision is challenged in light of freedom of expression and the right to vote as protected by ss. 2(b) and 3 of the Canadian Charter of Rights and Freedoms.

I. Factual Background

66 The appellants Thomson Newspapers Company Limited and Southam Inc. brought an application under rule 14.05(3)(g.1) of the Ontario Rules of Civil Procedure, R.R.O. 1990, Reg. 194, for the following relief:

(a) a declaration that s. 322.1 of the Canada Elections Act is of no force and effect in that it violates ss. 2(b) and 3 of the Canadian Charter of Rights and Freedoms and is not justified under s. 1; and

(b) a declaration directing that no proceedings may be brought under s. 322.1 pursuant to s. 126 of the Criminal Code, R.S.C., 1985, c. C-46, or any other statute or law.

67 On May 15, 1995, Somers J. of the Ontario Court (General Division) denied the appellants’ application on the basis that the impugned provision, although violating freedom of expression, was nonetheless justified under s. 1 of the Charter. Somers J. found no violation of the right to vote. On August 19, 1996, the Ontario Court of Appeal (Catzman, Carthy and Charron JJ.A.) dismissed the appeal and agreed with Somers J. that the right to vote had not been violated and the infringement of freedom of expression was justified.

II. Relevant Constitutional and Statutory Provisions

68 Canadian Charter of Rights and Freedoms

1. The Canadian Charter of Rights and Freedoms guarantees the rights and freedoms set out in it subject only to such reasonable limits prescribed by law as can be demonstrably justified in a free and democratic society.

2. Everyone has the following fundamental freedoms:

(b) freedom of thought, belief, opinion and expression, including freedom of the press and other media of communication;

3. Every citizen of Canada has the right to vote in an election of members of the House of Commons or of a legislative assembly and to be qualified for membership therein.

Canada Elections Act, R.S.C., 1985, c. E-2

255. The Chief Electoral Officer shall appoint a Commissioner of Canada Elections, in this Act referred to as the “Commissioner”, whose duties, under the general supervision of the Chief Electoral Officer, shall be to ensure that the provisions of this Act are complied with and enforced.

256. (1) No prosecution for an offence under this Act or for an offence under section 126 of the Criminal Code in relation to anything that this Act forbids or requires to be done shall be instituted except with the prior consent in writing of the Commissioner.

322.1 No person shall broadcast, publish or disseminate the results of an opinion survey respecting how electors will vote at an election or respecting an election issue that would permit the identification of a political party or candidate from midnight the Friday before polling day until the close of all polling stations.

Criminal Code, R.S.C., 1985, c. C-46

126. (1) Every one who, without lawful excuse, contravenes an Act of Parliament by wilfully doing anything that it forbids or by wilfully omitting to do anything that it requires to be done is, unless a punishment is expressly provided by law, guilty of an indictable offence and liable to imprisonment for a term not exceeding two years.

III. Judicial History

A. Ontario Court (General Division) (1995), 1995 7062 (ON SC), 24 O.R. (3d) 109

69 In the course of a survey of the social science evidence concerning the nature of opinion polls, Somers J. found that they have a recognized impact on the electoral choice of individual voters. Although the precise extent of this impact was difficult to characterize, he also found that there was widespread concern amongst experts, the public and participants in the electoral process about the effect of opinion polls. This concern related to the influence of polls generally, as well as to the potential influence of inaccurate polls published late in an election campaign. Somers J. observed that there was disagreement in the evidence about the overall effect of polls on elections and that no consensus existed in favour of the ban imposed by s. 322.1. He considered the polling ban to be part of a wider package of restrictions in the Canada Elections Act whose general purpose was to ensure fair elections.

70 With respect to the scope of s. 322.1, Somers J. concluded that the polling ban does not cover internal communications between clients and pollsters, but is directed at the dissemination of poll results to the electorate. He also excluded from the scope of s. 322.1 “hamburger polls”, namely marketing ploys used to sell consumer goods which are matched with political parties or candidates. In Somers J.’s view, these polls are clearly not statistical representations of opinion in general, and therefore, do not qualify as “opinion surveys”. Finally, Somers J. considered whether s. 322.1 prohibits discussion of all poll results including results that were published prior to the commencement of the ban. He concluded, at p. 132, that it would be absurd to prohibit discussion of these earlier polls:

One of the purposes of the legislation is to provide a response time to those who consider themselves victims of misleading polls. One of the most effective ways of responding is to refer to previous polls and show how the new poll is inconsistent with or less accurate than earlier polls. If such discussion of earlier polls is barred then the right to full response contemplated by Parliament is seriously weakened.

71 On the constitutional issues, Somers J. found, as the respondent conceded, that s. 322.1 infringed s. 2(b) of the Charter. Moreover, with regard to the type of expression at stake, Somers J. stated, at p. 135, that “even if polling is not, properly speaking, the political or partisan expression that is deemed most deserving of protection, it is by its very subject matter close to the core of s. 2(b)”. The right to vote contained in s. 3 of the Charter was found not to have been infringed. Somers J. wrote, at p. 137:

While the right to information gives substance to the right to vote, it remains ancillary to it. This means that under s. 3, the constitutional question is not directly whether the “right to information” was breached but whether a restriction placed on information has diminished or undermined the right to vote in a genuine election.

On the facts of this case, Somers J. concluded that the 72-hour polling ban was sufficiently short that it did not vitiate or truly undermine strategic voting based on poll results.

72 Somers J. found that s. 322.1 is a justifiable limit on freedom of expression under s. 1 of the Charter. He observed, at p. 142, that “even if social science evidence is found to be inconclusive, it is still possible to ground a pressing objective on evidence of extended public debate of an issue or by looking at the actions of other democracies”. The widespread concern about the effects of polls and the presence of polling bans in other democratic countries was, in Somers J.’s view, strong evidence of a pressing objective. A rational connection between the publication of polls and harm to the electoral process was also found. Parliament was said to have a “reasoned apprehension of harm” in the present case, following the test applied in R. v. Butler, 1992 124 (SCC), [1992] 1 S.C.R. 452. In Somers J.’s opinion, this apprehension of harm was based on: (a) the sheer prevalence of poll results; (b) the public’s awareness of poll results; (c) the failure to publish methodological information along with poll results; and (d) the potentially undue impact of polls published late in an election campaign. Further, Somers J. found that s. 322.1 minimally impairs freedom of expression and that alternative means — such as the mandatory publication of methodological information or measures designed to deter false polls — would be less effective. He added that these alternative measures would fail to adequately further one of Parliament’s objectives: to give the voter a period of rest and reflection prior to the vote. Finally, Somers J. indicated that the present case is similar to the situation in Irwin Toy Ltd. v. Quebec (Attorney General), 1989 87 (SCC), [1989] 1 S.C.R. 927, where Parliament had to balance the claims of competing groups and evaluate uncertain social science evidence. In Somers J.’s view, Parliament should be accorded a wide margin of flexibility in legislating in this area. The application was therefore dismissed with costs.

B. Ontario Court of Appeal (1996), 1996 890 (ON CA), 30 O.R. (3d) 350

73 The Court of Appeal disagreed with Somers J. regarding the ambit of s. 322.1. It held that the phrase “opinion surveys” should be read broadly to include not only scientific surveys, but also phone-in or mail-in polls, and even less scientific methods of gathering information, such as exit and “hamburger” polls. The Court of Appeal also disagreed with Somers J.’s conclusion that the prohibition in s. 322.1 is limited to new survey results. In the Court of Appeal’s view, s. 322.1 applies to publication or discussion of all opinion survey results, whether in the public domain prior to the commencement of the ban or not. There was agreement, however, with Somers J.’s view that internal or private communications are not caught by s. 322.1.

74 On the constitutional issues, the Court of Appeal was in substantial agreement with the analysis of Somers J. As to the character of the harm addressed by the publication ban, the Court of Appeal described the purpose somewhat more narrowly, at p. 353:

There is, however, no empirical evidence as to the extent or nature of the influence of opinion polls upon the voter, nor can it be said with certainty that the impact of opinion polls is undue. The real concern is that when opinion surveys are published as bare results, without methodological information, they have the potential to be deceiving, and even with such information they may require a response to explain their true significance.

75 The court found, at p. 359:

It is surely a substantial and pressing objective to respond to widespread perceptions that opinion surveys can be distorting and that response time is needed to avoid that danger, all in aid of creating a level playing field at that critical point in the electoral process when the candidates are about to be selected.

76 The Court of Appeal also agreed with Somers J. that the ban on opinion survey results does not violate the right to vote under s. 3 of the Charter. The court held, at p. 358:

We accept that, as a general principle, the right to cast an informed ballot carries with it the right to information necessary to permit electors to vote rationally and in an informed manner. . . . The right to cast an informed ballot does not, in our view, elevate the provision of a snapshot of the mood of the electorate at a particular time to the level of a constitutional entitlement during the last three days of an election campaign.

77 The Court of Appeal found the infringement of the freedom of expression to be justified under s. 1 of the Charter. The court stated, at p. 360:

There was serious controversy on a social scientific subject as to the effect of polls upon the electorate, combined with the manifest fact that the publication of bare results does not tell the whole story and thus may well be misleading. This constitutes a reasonable apprehension of harm and the rational connection is then further exhibited by directing the legislation to the final three days of the election.

Even the broader interpretation of s. 322.1 favoured by the Court of Appeal was found to pass the minimal impairment analysis. In the result, the appeal was dismissed with costs.

IV. Issues

78 The following constitutional questions were stated by the Chief Justice on March 27, 1997:

1. Does s. 322.1 of the Canada Elections Act, R.S.C., 1985, c. E-2, as amended, infringe s. 2(b) and/or s. 3 of the Canadian Charter of Rights and Freedoms?

2. If s. 322.1 of the Canada Elections Act infringes s. 2(b) and/or s. 3 of the Canadian Charter of Rights and Freedoms, is s. 322.1 a reasonable limit prescribed by law as can be demonstrably justified in a free and democratic society for the purposes of s. 1 of the Canadian Charter of Rights and Freedoms?

V. The Right to Vote

79 I find it necessary, at the outset of my analysis on the right to vote, to distinguish between the two Charter rights at issue in the present case. It is significant, for instance, that s. 3 of the Charter, which guarantees the citizen’s right to vote, is not subject to override under s. 33 of the Charter. This means that a statutory provision which violates s. 3, and is not saved by s. 1, cannot be insulated from Charter review by Parliament or a provincial legislature. By contrast, s. 2(b) of the Charter, which protects free expression, is subject to override under s. 33. Even though the override power is rarely invoked, the fact that s. 3 is immune from such power clearly places it at the heart of our constitutional democracy.

80 Moreover, in cases where freedom of expression and the right to vote may overlap or come into conflict, it is necessary to find an appropriate balance between both sets of rights. Support for this conclusion may be found in Dagenais v. Canadian Broadcasting Corp., 1994 39 (SCC), [1994] 3 S.C.R. 835. In that decision, this Court considered the balance to be achieved between the right to a fair trial and freedom of expression, in the context of publication bans. Lamer C.J. wrote for the majority, at p. 877:

A hierarchical approach to rights, which places some over others, must be avoided, both when interpreting the Charter and when developing the common law. When the protected rights of two individuals come into conflict, as can occur in the case of publication bans, Charter principles require a balance to be achieved that fully respects the importance of both sets of rights.

In my view, these comments are equally applicable where the right to vote overlaps with the right to free expression. Each right is distinct and must be given effect.

81 It is noteworthy, as well, that the scope of s. 2(b), unlike the scope of s. 3, has been well canvassed by this Court. Most recently, the broad interpretation of the scope of s. 2(b) was affirmed in Libman v. Quebec (Attorney General), 1997 326 (SCC), [1997] 3 S.C.R. 569. There, the Court wrote, at para. 31: “Unless the expression is communicated in a manner that excludes the protection, such as violence, the Court recognizes that any activity or communication that conveys or attempts to convey meaning is covered by the guarantee of s. 2(b) of the Canadian Charter”.

82 Turning to s. 3, this Court has only dealt with the scope of the right to vote in limited contexts, namely electoral boundaries, the disqualification of inmates from voting, and voting in a referendum; see Reference re Provincial Electoral Boundaries (Sask.), 1991 61 (SCC), [1991] 2 S.C.R. 158; Sauvé v. Canada (Attorney General), 1993 92 (SCC), [1993] 2 S.C.R. 438; and Haig v. Canada, 1993 58 (SCC), [1993] 2 S.C.R. 995. This Court has not addressed the scope of s. 3 in the context of access to information during an election. In my view, it is not necessary to decide on the informational content of s. 3 in the context of this case. The purpose of s. 3 was stated clearly by McLachlin J. in Reference re Provincial Electoral Boundaries (Sask.), supra, at p. 183, as the “right to ‘effective representation’”; see also Haig, supra, at p. 1031. Accordingly, to constitute an infringement of the right to vote, a restriction on information would have to undermine the guarantee of effective representation.

83 This position accords with the jurisprudence of the European Court of Human Rights and the European Commission of Human Rights. Article 3 of the First Protocol to the European Convention for the Protection of Human Rights and Fundamental Freedoms, March 20, 1952, Europ. T.S. No. 9, provides that parties to the Convention “undertake to hold free elections at reasonable intervals by secret ballot, under conditions which will ensure the free expression of the opinion of the people in the choice of the legislature”. The European Court has held that this provision guarantees the right to vote; see Eur. Court H.R., Mathieu-Mohin and Clerfayt case, judgment of 2 March 1987, Series A No. 113. However, neither the European Court nor the European Commission has equated the right to vote with a right to information per se. Rather, in Mathieu-Mohin and Clerfayt, supra, the European Court equated the right to vote with the right to participate in the electoral process. This same principle was applied by the European Commission in Bowman v. United Kingdom (1996), 22 E.H.R.R. C.D. 13. In that case, the Commission considered legislation that restricted “single-issue” campaigning by individuals other than the electoral candidates. The applicant, who had been charged for distributing leaflets outlining the views of three electoral candidates on abortion, complained that her right to free expression had been violated. The Commission agreed and held that there was an unjustified violation of freedom of expression. With regard to the applicant’s right to vote, the Commission wrote, at p. CD18:

The Commission has had regard to whether the expression of opinion or information on “single issues” addressed by individuals or groups with strongly-held views may operate in particular constituencies so as to “distort” election results. It has previously considered in the context of Article 3 of the First Protocol . . . that one of the legitimate objectives of national electoral systems is to channel currents of thought so as to promote the emergence of a sufficiently clear and coherent political will. . . . The Government has not, however, produced any argument to the effect that “single issue” campaigning of the kind illustrated in the applicant’s case would distract voters from the political platforms which are the basis of national party campaigns to such a degree as would hinder the electoral process. [Emphasis added.]

84 In the present instance, it was argued that the partial ban on election polls hinders the electoral process because some of the voters are deprived of information relevant to the exercise of their right to vote. The Court of Appeal considered, at p. 358, that s. 3 only guarantees access to information “necessary to permit electors to vote rationally and in an informed manner”, and that poll results did not fall into that category. My conclusion that s. 322.1 is an unjustified limit on free expression makes it unnecessary to deal with this issue. I prefer to leave the issue aside mostly because the evidence with regard to the relationship between the kind of information banned and the integrity of the election process is too sparse.

VI. Freedom of Expression

A. Section 2(b) of the Charter

85 Before this Court, the respondent conceded, as it did in the courts below, that there was a prima facie infringement of freedom of expression. This concession was well-founded. Section 322.1 clearly infringes on the guarantee in s. 2(b) of the Charter, in accordance with the test set out in Irwin Toy, supra. First, there can be no doubt that the publication of polling information, and more specifically opinion survey results, is an activity that conveys meaning and, therefore, falls within the ambit of s. 2(b). Second, s. 322.1 restricts freedom of expression by prohibiting the broadcasting, publication or dissemination of opinion survey results during the final three days of an election campaign. The freedom of expression is clearly infringed by this ban.

86 Before turning to the justification analysis under s. 1 of the Charter, I pause to note that, in addition to polling, there are various other activities subject to restriction during a federal election. These restrictions, as set out in the Canada Elections Act, include limits on spending and controls on advertising. In my view, it is impossible to make any generalizations regarding the scope or constitutional validity of these provisions. Each type of restriction, including s. 322.1, must be considered in its own context. Although Somers J. noted that the partial ban on polling is part of a larger package of restrictions, there is no justification for finding that all restrictions must be considered globally. Each restriction is adopted to address a particular set of circumstances and it is in that light that each one must be analysed.

B. Justification Under Section 1 of the Charter

1. Contextual Factors

87 The analysis under s. 1 of the Charter must be undertaken with a close attention to context. This is inevitable as the test devised in R. v. Oakes, 1986 46 (SCC), [1986] 1 S.C.R. 103, requires a court to establish the objective of the impugned provision, which can only be accomplished by canvassing the nature of the social problem which it addresses. Similarly, the proportionality of the means used to fulfil the pressing and substantial objective can only be evaluated through a close attention to detail and factual setting. In essence, context is the indispensable handmaiden to the proper characterization of the objective of the impugned provision, to determining whether that objective is justified, and to weighing whether the means used are sufficiently closely related to the valid objective so as to justify an infringement of a Charter right.

88 Characterizing the context of the impugned provision is also important in order to determine the type of proof which a court can demand of the legislator to justify its measures under s. 1. This question is suitably addressed at the outset because it affects the entirety of the s. 1 analysis, and because of the nature of the evidence in this case. In discussing whether a ban on advertising aimed at children minimally impaired the freedom of expression, Dickson C.J. and Lamer and Wilson JJ. observed in Irwin Toy, supra, at pp. 993-94:

When striking a balance between the claims of competing groups, the choice of means, like the choice of ends, frequently will require an assessment of conflicting scientific evidence and differing justified demands on scarce resources. . . . Thus, as courts review the results of the legislature’s deliberations, particularly with respect to the protection of vulnerable groups, they must be mindful of the legislature’s representative function. . . .

In other cases, however, rather than mediating between different groups, the government is best characterized as the singular antagonist of the individual whose right has been infringed. . . .

In the instant case, the Court is called upon to assess competing social science evidence respecting the appropriate means for addressing the problem of children’s advertising. The question is whether the government had a reasonable basis, on the evidence tendered, for concluding that the ban on all advertising directed at children impaired freedom of expression as little as possible given the government’s pressing and substantial objective. [Emphasis added.]

In Butler, supra, Sopinka J. found the social science evidence relating pornography to violence or other harms directed at women by men to be inconclusive. Nonetheless, he found that there was a rational connection between the impugned provision and the measures adopted by the legislature (at pp. 502 and 504):

While a direct link between obscenity and harm to society may be difficult, if not impossible, to establish, it is reasonable to presume that exposure to images bears a causal relationship to changes in attitudes and beliefs.

I am in agreement with . . . the view that Parliament was entitled to have a “reasoned apprehension of harm” resulting from the desensitization of individuals exposed to materials which depict violence, cruelty, and dehumanization in sexual relations. [Emphasis added.]

In RJR-MacDonald Inc. v. Canada (Attorney General), 1995 64 (SCC), [1995] 3 S.C.R. 199, McLachlin J. addressed more explicitly the relationship between the standard of proof required under s. 1 and the nature of the problem which Parliament was seeking to remedy. That case involved the application of s. 2(b) to certain restraints on tobacco advertising and promotion which had been adopted by Parliament. After noting that the distinction described in Irwin Toy between the government as social mediator and the government as singular antagonist “may not always be easy to apply”, she observed, at para. 135:

This said, I accept that the situation which the law is attempting to redress may affect the degree of deference which the court should accord to Parliament’s choice. The difficulty of devising legislative solutions to social problems which may be only incompletely understood may also affect the degree of deference that the courts accord to Parliament or the Legislature.

She emphasized, however, that this does not diminish the usual standard of proof required under s. 1, simply that that standard might be satisfied in different ways depending on the nature of the legislative objective (at para. 137):

As the s. 1 jurisprudence has established, the civil standard of proof on a balance of probabilities at all stages of the proportionality analysis is more appropriate: Oakes, supra, at p. 137; Irwin Toy, supra, at p. 992. . . . Discharge of the civil standard does not require scientific demonstration; the balance of probabilities may be established by the application of common sense to what is known, even though what is known may be deficient from a scientific point of view: see Snell v. Farrell, 1990 70 (SCC), [1990] 2 S.C.R. 311. [Emphasis added.]

McLachlin J. applied this test under the rational connection stage of the proportionality analysis based on the following characterization (at para. 154):

Where, however, legislation is directed at changing human behaviour, as in the case of the Tobacco Products Control Act, the causal relationship may not be scientifically measurable. In such cases, this Court has been prepared to find a causal connection between the infringement and benefit sought on the basis of reason or logic, without insisting on direct proof of a relationship between the infringing measure and the legislative objective: R. v. Keegstra, 1990 24 (SCC), [1990] 3 S.C.R. 697, at pp. 768 and 777; R. v. Butler, 1992 124 (SCC), [1992] 1 S.C.R. 452, at p. 503. [Emphasis added.]

89 The issue of the standard and type of proof required under s. 1 arises with particular acuity in the case at bar because the social science evidence, as in these previous cases, is in a state of some controversy. In light of this inconclusive evidence, the government submitted it is not for this Court to second-guess the judgment of the legislature when it has made a reasonable assessment that an apprehension of harm exists. It also argued that “common sense applied to what is known establishes the reasonableness of Parliament’s assessment of the situation”.

90 I agree with McLachlin J.’s remarks in RJR-MacDonald that it is difficult to draw a sharp distinction between legislation in which the state is the antagonist of the individual, and that in which it is acting as a mediator between different groups. Indeed, nothing in these cases suggests that there is one category of cases in which a low standard of justification under s. 1 is applied, and another category in which a higher standard is applied. In my view, these cases further the contextual approach to s. 1 by indicating that the vulnerability of the group which the legislator seeks to protect (as in Irwin Toy, supra, at p. 995; Ross v. New Brunswick School District No. 15, 1996 237 (SCC), [1996] 1 S.C.R. 825, at para. 88), that group’s own subjective fears and apprehension of harm (as in R. v. Keegstra, 1990 24 (SCC), [1990] 3 S.C.R. 697, per McLachlin J., at p. 857), and the inability to measure scientifically a particular harm in question, or the efficaciousness of a remedy (as in Butler, supra, at p. 502), are all factors of which the court must take account in assessing whether a limit has been demonstrably justified according to the civil standard of proof. They do not represent categories of standard of proof which the government must satisfy, but are rather factors which go to the question of whether there has been a demonstrable justification. I propose to return to these factors in more detail in the course of a contextual approach to s. 1.

91 Another contextual factor to be considered is the nature of the activity which is infringed. The degree of constitutional protection may vary depending on the nature of the expression at issue (Edmonton Journal v. Alberta (Attorney General), 1989 20 (SCC), [1989] 2 S.C.R. 1326, at pp. 1355‑56; Rocket v. Royal College of Dental Surgeons of Ontario, 1990 121 (SCC), [1990] 2 S.C.R. 232, at pp. 246‑47; Keegstra, supra, at p. 760; RJR‑MacDonald, supra, at paras. 71-73 and 132; Libman, supra, at para. 60). This is not because a lower standard is applied, but because the low value of the expression may be more easily outweighed by the government objective. In this case, the speech infringed is political information. While opinion polls may not be the same as political ideas, they are nevertheless an important part of the political discourse, as manifested by the attention such polls receive in the media and in the public at large, and by the fact that political parties themselves purchase and use such information. Indeed, the government argues that opinion polls have an excessive impact on the electoral choices made by voters. As a genre of speech, unlike hate speech or pornography, this expression is not intrinsically harmful or demeaning to certain members of society because of its direct impact, or its impact on others. It is without moral content, and yet it is widely perceived as a valuable and important part of the discourse of elections in this country. The government urges, however, that under some circumstances polls may come to have an effect which interferes with the ability of individuals to make an informed choice.

92 According to the purposes I describe below, there are two groups who might be negatively affected by polls: first, there are those who incorrectly assume that polls are a perfect measure of voting results on election day, and rely on them to an excessive degree in consequence; second, there are those voters who are perfectly aware of the general shortcomings of polls as predictions of the result on election day, but who are misled by the publication of an inaccurate poll result. Even assuming there to be some likelihood of these dangers, which shall be discussed more fully in the course of the s. 1 analysis proper, there can be no question that opinion surveys regarding political candidates or electoral issues are part of the political process and, thus, at the core of expression guaranteed by the Charter. As Dickson C.J. stated in Keegstra, supra, at pp. 763-64:

The connection between freedom of expression and the political process is perhaps the linchpin of the s. 2(b) guarantee, and the nature of this connection is largely derived from the Canadian commitment to democracy. Freedom of expression is a crucial aspect of the democratic commitment, not merely because it permits the best policies to be chosen from among a wide array of proffered options, but additionally because it helps to ensure that participation in the political process is open to all persons.

In that case, hate speech was found to interfere with the ability of a specific and identifiable group to participate in the political process by directly undermining their dignity and their membership in the community. The same could be said of pornographic expression in Butler. And in Irwin Toy, the interest of advertisers meant that there was a likelihood that such speech would be manipulative of children and would play on their vulnerability.

93 In each of these cases, the type of speech involved systematically and consistently undermined the position of some members of society. There is no evidence in this case that there is any such systematic opposition between the interests or position of the Canadian voter and opinion surveys. The government argues that there is the potential that some inaccurate poll might undermine the freedom of choice of the Canadian voter, or that some voters might be excessively influenced by polls. Leaving those exceptional or potential cases aside, polls are not generally inimical to the interests of Canadian voters. They are sought after and widely valued which, independently of their value to any one voter or specific content, places this type of speech at the core of the political process.

94 Although the Libman case involved expression related to political campaigning, there was a likelihood that the genre of paid political advertising would significantly manipulate the political discourse to the advantage of those with greater financial resources (paras. 50-51). Libman is not dissimilar to Irwin Toy in the sense that, under certain circumstances, the nature of the interests (i.e., a single party or faction with a great preponderance of financial resources) of the speakers could make the expression itself inimical to the exercise of a free and informed choice by others. The government does not suggest that there is any such systematic or structural danger in the case of opinion surveys, but relies simply on the possibility of an inaccurate poll, or the disproportionate reaction by certain voters to polls generally. These may be important objectives and will be assessed below; but the possibility of harm arising from the unfortuitous publication of an inaccurate poll does not displace the general nature of this expression as political expression at the core of s. 2(b).

95 I would conclude that the nature of the expression in issue here does not prima facie suggest that a deferential approach is appropriate in this case.

2. Legislative Objective

96 The characterization of the objective of s. 322.1 and the nature of the harm which it hopes to remedy are virtually correlatives of one another, and constitute the crucial first step in the s. 1 analysis. Unfortunately, the decisions below and some of the submissions before us have been unnecessarily vague in stating the precise objective of the limitation in issue, leaving many unanswered questions as to the proper approach to the s. 1 analysis. The respondent submitted before us that the objective of s. 322.1 is “to prevent the distorting effect of public opinion survey results which are released late in the election when there is no longer a sufficient opportunity to respond”. But what is the exact nature of this “distorting effect”? The Court of Appeal did not elucidate any more precise objective than this. By contrast, Somers J. articulated four objectives from the legislative history of the provision which more precisely characterized the purpose of the limit on freedom of expression: prevention of dissemination of false information; prevention of polling information from being presented in a misleading way that lent an aura of scientific precision to poll results; a period of rest and reflection for voters prior to going to the polls; and a period of response after the final poll has been published, presumably to respond to its potential inaccuracy or simply to question the importance of polls in general. Before this Court, the respondent submitted that the sole objective of the provision on which it was relying was to provide a period during which the accuracy of a poll could be publicly questioned and debated, so that the scientific validity and accuracy of any particular poll would be more fully known to Canadian voters. The harm, then, which the provision purportedly addresses is the possibility that the scientific accuracy of polls may be overestimated by Canadian voters in a particular election and that they may cast their vote based on this inaccurate perception.

97 As mentioned in the previous section, the voter’s misapprehension of the true significance of a poll could be the result of either of two quite different reasons: first, the voter might systematically overestimate the validity and accuracy of poll results; or, second, there might be a poll which falls below the normal standard of accuracy of polling which Canadians are generally entitled to expect. This latter type of poll is more simply described as a bad poll, a false poll, or an inaccurate poll. The respondent is not explicit as to which distorting effect it refers, and the judgments below, as well as legislative pronouncements at the time of its passage, reflect this obscurity. The danger of the systematically overreliant voter is suggested in the purpose described by Somers J. as the need for “a rest period” so that the frenzy of polls will die down and voters will be encouraged to forget about the polls and concentrate on issues. This purpose is evident in the final report of the Royal Commission on Electoral Reform and Party Financing (“Lortie Commission”) entitled Reforming Electoral Democracy (1991), vol. 1, which concluded, at p. 460, that a polling ban was necessary to “provide voters with a period of reflection at the end of campaign to assess the parties and candidates”. The second purpose of the legislation, to guard against an inaccurate poll which occurs late in the campaign, has nothing to do with a period of repose or reflection. Rather, the purpose is to provide an opportunity for the last opinion surveys on which the voter might base his or her vote to be subjected to public scrutiny. In this way, the voter will have the best information possible about the accuracy of the latest polls, and will not cast a vote without a potentially inaccurate opinion survey having been publicly scrutinized.

98 At a higher level of generality, the purpose of providing more accurate information to Canadian voters is that they are more capable of making a free and informed choice, which engenders a freer and fairer election process. It is said that the election process will also be perceived as fairer by the electorate with this restriction in place. This, in turn, strengthens democracy. For the purpose of the s. 1 analysis, however, it is desirable to state the purpose of the limiting provision as precisely and specifically as possible so as to provide a clear framework for evaluating its importance, and the precision with which the means have been crafted to fulfil that objective (RJR-MacDonald, supra, at para. 144; Vriend v. Alberta, 1998 816 (SCC), [1998] 1 S.C.R. 493, at para. 110).

99 Accepting these as the two objectives of s. 322.1, it is my view that any ambiguity in the words of the section should be interpreted in accordance with those purposes, rather than to frustrate them. I conclude, therefore, that Somers J. was correct in interpreting the prohibition in s. 322.1 as applying only to “new” poll results, i.e. results that are undisclosed as of midnight on the Friday before election day. Far from preventing old poll results from being mentioned, the very raison d’être of the section is that those old poll results should be aired and discussed in the media so that their accuracy can be fully determined in public debate.

3. Is the Objective Pressing and Substantial?

100 For clarity, I propose to examine each purpose separately, beginning with the “period of rest” purpose. I will examine this purpose briefly as the respondent all but withdrew this as part of its argument before this Court. However, I still consider this analysis necessary as the vague references to the “distorting effect” of polls could still accommodate this purpose. Moreover, there is little doubt that this was one of the original purposes of Parliament in enacting the provision. In addition to the statement in the final report of the Lortie Commission mentioned above, Senator Rivest, in introducing the bill containing s. 322.1 in the Upper House, declared that it was an attempt to [translation] “reconcile the freedom of speech and the freedom of the press with the right of the voters to make a judgement peacefully, come election time”: Debates of the Senate, April 29, 1993, at p. 3117.

101 An examination of this purpose reveals some disturbing assumptions. First, this purpose does not rely on the inaccuracy of any opinion survey results. Rather, it suggests that Canadians will become so mesmerized by the flurry of polls appearing in the media that they will forget the issues upon which they should actually be concentrating. This reasoning cannot be countenanced. Canadian voters must be presumed to have a certain degree of maturity and intelligence. They have the right to consider the results of polls as part of a strategic exercise of their vote. It cannot be assumed that in so doing they will be so naïve as to forget the issues and interests which motivate them to vote for a particular candidate. Nor can Canadians be presumed to assume that polls are absolutely accurate in predicting outcomes of elections and that they thus will overvalue poll results. Many polls are released in the course of an election campaign which belies the suggestion that any one poll could be perceived as authoritative. These opinion polls yield differing results even when conducted contemporaneously, and, perhaps more importantly, opinion poll results fluctuate dramatically over time. I cannot accept, without gravely insulting the Canadian voter, that there is any likelihood that an individual would be so enthralled by a particular poll result as to allow his or her electoral judgment to be ruled by it.

102 I am thus unable to perceive, and nor has the government seriously argued before us, that any pressing and substantial objective is served by the existence of a “rest period” for polls prior to the election date. I would, therefore, find that s. 322.1 is not justified under s. 1 according to this objective.

103 The more difficult question is whether the blackout period can be justified by the other legislative purpose, which is to prevent an inaccurate poll from having an impact on voter choice because of a lack of response time prior to the voter casting his or her ballot. The first step in determining whether this is a pressing and substantial objective is to determine whether such a poll would actually influence voter choice, and to what extent; the second step is to evaluate the likelihood of an inaccurate poll being published.

104 Although the extent of the influence of polls on voter choice is uncertain, there is evidence suggesting that it may be significant. The key evidence in this case was gathered by the Lortie Commission and presented in its final report. The Lortie Commission, supra, at p. 457, concluded that: “Notwithstanding the frequent assertion of pollsters that their data have minimal influence on voters, recent research provides strong support for the proposition that published opinion polls can significantly influence campaigns and voters”. More specifically, the Commission asserted that “the argument that published polls do not influence voter choice or affect the conduct of campaigns is simply untenable” (p. 458). Thus, the Commission found that polls not only have a general impact on the conduct of an election, but that they affect voter choice. The harmful influence, according to the Commission, arose because opinion polls are subject to errors which are not fully disclosed to the public (at p. 455):

Because they are presented as ‘scientific’, published opinion polls raise issues of public confidence in the integrity of the electoral process. Notwithstanding their claims to scientific validity and accuracy in representing the views of all potential voters, opinion polls are susceptible to many forms of error and misrepresentation. The apparent precision of the data they report fails to reflect the fact that they are estimates of the distribution of opinion at a given time. Yet their apparent authority gives them considerable influence over the conduct of campaigns and the choices made by voters. [Emphasis added.]

Again, I would observe that there appear to be two strains of thought in this passage: first, that the accuracy of any poll may be overestimated by Canadians, and second, that some polls may fall below the standard of accuracy generally expected through error or misrepresentation. I have found above that the first of these objectives is not of pressing and substantial concern justified under s. 1. However, the Lortie Commission appears also to be concerned about the influence of an inaccurate poll. The Lachapelle Study (Polls and the Media in Canadian Elections: Taking the Pulse (1991)) and the White Paper on Election Law Reform (1986) were considerably less certain of this influence, however. Drawing on the evidence presented to the Lortie Commission, Professor Lachapelle was able to identify a number of discrete influences of opinion polls, including: the bandwagon effect (rallying to support the leader in the polls); the underdog effect (rallying to support the trailing candidate); the demotivating effect (abstaining from voting); the motivating effect (encouraging voters to cast their ballots); the strategic effect (electors decide how to vote on the basis of the relative popularity of parties); and the free-will effect (voting to prove the polls wrong); see Lachapelle Study, supra, at pp. 13-14. There is dispute, however, regarding the overall influence on election results. For instance, it is suggested that one effect is simply counterbalanced by another and, therefore, the overall influence of polls is nil; see White Paper, supra, at p. 26. Professor Lachapelle cautioned that “no definitive conclusion about the actual impact of polls can be drawn from these briefs” (p. 14), and that “theories about the decline of political parties and the undue influence of polls on voters have not been convincingly supported” (p. 29).

105 Although the overall influence of polls may not have been scientifically established as “undue”, I would nevertheless conclude that there is evidence of significant influence of polls on the electoral process and on individual electoral choice. Although the overall effect of these polls may be difficult to discern or predict, this evidence suggests that an uncertain number of voters might be influenced in their electoral choice by this false information. The pernicious aspect of an inaccurate poll is that no voter could discover its true nature because of the lack of response time.

106 The possibility of the publication of an inaccurate poll is not de minimis. The Lortie Commission, supra, at p. 457, reported that:

Although the industry in general has become highly professional since public polling was introduced in Canada in 1941, the incidence of technically deficient and poorly reported polls is still substantial. In recent elections, there have been instances of misleading polls, some because of technical errors and others because of partisan misrepresentation. There have even been allegations of fraudulent polls, where the data were said to have been fabricated to counter a poll showing the opposition in the lead. Such “bogus” polls and the more common misrepresented poll have been released to the media in many democracies. (Cantril 1991, 67; Worcester 1991, 199; Hoy 1989, 189-202) It is the willingness of the media to report such polls that makes them significant and troublesome.

The close relationship between some polling organizations and political parties also suggests that polls released as purportedly scientific measures of public opinion could be subject to manipulation (Lachapelle Study, supra, at p. 133).

107 There is also evidence of public and governmental concern to guard against inaccurate polls which are published late in the campaign and which thus cannot be subject to the same scrutiny as polls published earlier in the election. The respondent submitted the results of a poll indicating that 45 percent of Canadians are in favour of such a ban, 28 percent are opposed, and 27 percent had no opinion. As a simple matter of logic, moreover, clear evidence of the influence of polls on individual voter choice, combined with indications that such inaccurate polls are not a remote possibility, suggests that the voting decision of those who rely on polls as part of their decision-making process could be distorted. The possibility of such a distortion is clearly a matter which the government may legitimately be concerned to remedy.

108 The validity of this concern is attenuated, however, by an important disjunction between the evidence and the harm which the legislation purports to address. The evidence and conclusions presented by the Lortie Commission and the Lachapelle Study relate to the influence of polls in the aggregate; the purported objective of this legislation relates to the inaccuracy of an individual poll which undeservedly benefits from the perception of scientific accuracy as a result of the generally high standards of accuracy maintained by polls, and by its presentation in the media as a scientific measure. But an opinion poll does not appear in a vacuum. Rather, it is published chronologically after a series of other polls which have been measuring public opinion throughout the election. In all likelihood, other polls conducted by other polling organizations will appear in other media outlets during the three days prior to election day. Thus, to the extent that any single poll is inaccurate, this will possibly be apparent to voters who are aware of the results of other polls, both those published immediately prior to this final period before the election, and those appearing in the media at the same time as the inaccurate poll. The more polls which appear during this period, the less likely that voters will base their decisions on the inaccurate poll. Moreover, the severity of the error, which one might speculate might enhance the influence of the error, would be offset by increased ease with which the mistake would be identified. In addition, voters’ experience with opinion polls in previous elections will have demonstrated that opinion polls are of variable value and accuracy as predictive measures of the outcome of an election.

109 Notwithstanding this qualification, I conclude that the purpose of guarding against the possible influence of inaccurate polls late in the election campaign by allowing for a period of criticism and scrutiny immediately prior to election day, is a pressing and substantial objective. As this Court stated in Harvey v. New Brunswick (Attorney General), 1996 163 (SCC), [1996] 2 S.C.R. 876, at para. 38, measures that “maintain and enhance the integrity of the electoral process . . . [are] always of pressing and substantial concern in any society that purports to operate in accordance with the tenets of a free and democratic society”. The purpose of this particular limitation on expression is to ensure that information which the evidence indicates has an important influence on the choice of at least some voters is presented according to the standards of accuracy which polls are normally expected to attain. When pollsters and the media present information which aspires to certain scientific standards of quality, and which invites reliance by voters in the exercise of their vote, then the government may legitimately be concerned. Such information is qualitatively different from partisan rhetoric, or even journalistic reporting which aspires to certain standards of accuracy and objectivity. Polling information is presented scientifically which reflects relatively settled and defined standards for determining accuracy. To the extent that the votes of some might be distorted as a result of polls being presented in a misleading fashion, this is a pressing and substantial objective.

4. Rational Connection

110 The three-day blackout period on the publication of polls will serve, to some degree, the purpose of preventing the use of inaccurate polls by voters. The blackout period gives critics the opportunity to assess the methodological information made available by the pollster and to question the validity of the poll on that basis. To that extent, the ban is rationally connected to the purpose of the legislation. However, s. 322.1 does not prevent an opinion survey from being released without any information as to methodology. Thus, there may be cases where it would be impossible for outside observers to scrutinize or challenge the validity of a poll. In those cases, the most that could be achieved by the blackout period is that the validity of a poll could be undermined by pointing out the failure of the pollster to publish the methodology of the poll. Having mentioned this infirmity in the connection between the purpose of the provision and the means designed to carry out that purpose, I prefer to focus my analysis of the inadequacies of this legislation under the rubric of minimal impairment.

5. Minimal Impairment

111 Section 322.1 does not minimally impair the right to freedom of expression guaranteed in the Charter and is, therefore, not justified under s. 1. Indeed, it is my view that s. 322.1 is a very crude instrument in serving the purpose articulated by the government in this case. To repeat, that objective is to prevent or minimize the distorting effect of inaccurate opinion polls, and, in particular, opinion polls released late in an election which may have an undue influence on voters and which are not subject to adequate criticism prior to the voter relying on them. As I have stated at the outset, the type of proof required to discharge the burden of justification on the government may vary from case to case depending on the context. In this section, I address a number of contextual factors pertaining to the seriousness and likelihood of the harm, as well as the standard and methods of proof in a case such as this one involving the evaluation of social science evidence and human behaviour. These contextual factors bear on the degree of deference which a court should accord to the particular means chosen to implement a legislative purpose; see RJR-MacDonald, supra, at paras. 132 and 160.

112 The first factor which could militate in favour of a deferential approach towards the legislation is the vulnerability of the group sought to be protected. In this case, however, the social science evidence did not establish that the Canadian voter is a vulnerable group relative to pollsters and the media who publish polls. The presumption in this Court should be that the Canadian voter is a rational actor who can learn from experience and make independent judgments about the value of particular sources of electoral information. As Professor Lachapelle’s Study indicates, some voters clearly do consider polls to be of some value in making their electoral decision (supra, at p. 13). However, no evidence has been presented before this Court that voters have suffered from any misapprehensions regarding the accuracy of any single poll. Indeed, the fact that polls conducted contemporaneously yield differing results, or that poll results can fluctuate dramatically over time, suggests that voters have experience with the shortcomings of some polls. Indeed, perhaps the most important contextual factor in the analysis of this case is that polls are widely available throughout the election period in large numbers and from differing media sources. Professor Lachapelle reports, supra, at p. 86, that a total of 59 polls were published during the 1988 national election campaign, 22 of them national in scope, and 37 regional. Voters are constantly exposed to opinion poll results throughout the election and a single inaccurate poll result is likely to be spotted and discounted appropriately.

113 The government responded to the paucity of evidence relating to this specific issue by stating that it was sufficient “to show a reasoned apprehension of harm. . . . [I]t is sufficient if it is reasonable to presume that there is a casual [sic] relationship between them”. It relied on a number of cases for the proposition that even in the face of inconclusive social science evidence, the Court has adopted a deferential approach to determining whether the harm exists and in assessing the justification of the measures chosen to prevent those harms: Butler, supra, RJR-MacDonald, supra, Libman, supra, Keegstra, supra, and Irwin Toy, supra. In my view, the principles developed in those cases are not applicable here for three fundamental reasons. First, the presumptions which the Court made in those cases were not refuted by any contrary logical reasoning. That is not the case here. What I have said in the previous paragraph suggests, as a matter of logic, that there is reason to believe that, notwithstanding the scientific “aura” of polls, the Canadian voter is likely to be aware of a seriously inaccurate poll. Indeed, the more serious the inaccuracy, the more likely the awareness of the error. This is not to say, of course, that some voters might not be misled by an inaccurate poll and cast their vote on what amounts to a misrepresentation. Indeed, that possibility is precisely what I have found constitutes the pressing and substantial objective of this provision. The point here is simply that the claims of widespread or significant harm based on logical inferences derived from surrounding factors are not compelling in the context of factors which refute such logical inferences. As a matter of logical reasoning and inference, I find the government’s claims of harm to be controverted by surrounding circumstances which suggest that the Canadian voter is already sensitive to the danger which the government is seeking to remedy.

114 Second, there is no suggestion in this case that the interests of the voter and of the pollster are opposed, or that the latter will systematically attempt to manipulate the former. This sets this case apart from cases involving advertisers (such as Irwin Toy, Libman or RJR-MacDonald), where advertisers encouraged choices which served their particular interests. Although it was legal for each of them to pursue those interests, and therefore legitimate to do so, there was a danger of undue manipulation in the first two cases, and of serious health consequences for individual Canadians in the third. Thus, not only was the government dealing with a situation in which expression would be used to manipulate vulnerable groups, but also in which there was a balancing of conflicting but legitimate social interests. None of those elements are present in the case at bar. There is no interest which favours inaccurate or misleading polls being foisted on the Canadian public. The media have an interest in providing poll results which are of interest to Canadians and which uphold their reputation for integrity and accuracy. Pollsters have an interest in maintaining a reputation for accurately measuring public opinion and election results. Although some individual polling firms have close relationships with particular parties which might encourage them, consciously or unconsciously, to tailor their methodologies in favour of their valued client, the polling industry as a whole does not favour one party over another (Lachapelle Study, supra, at p. 133). To the extent that individual polls may reflect a bias, divergent opinion survey results reflecting these biases will be obvious to voters and diminish the scientific aura of those polls. Voters, the third interested group, also desire accurate polls so that if they choose to rely on them, the most accurate and timely polls will be available. Unlike the advertising cases, this is not a case in which the government is intervening against a powerful interest to prevent expression from being a means of manipulation and oppression.

115 Third, the reasonable apprehension of harm test has been applied where it has been suggested, though not proven, that the very nature of the expression in question undermines the position of groups or individuals as equal participants in society. This has been accepted, in particular, when it is difficult or impossible to establish scientifically the type of harm in question. The respondent relied particularly on the reasoning in Butler, supra, to justify the reasoned apprehension of harm standard. In my view, the reasoning used in that case is not applicable here. In Butler, the difficulty of scientifically proving the harm in question was not the only feature of the case justifying the application of the reasonable apprehension of harm test. First, there was, in that case, “a substantial body of opinion that holds that the portrayal of persons being subjected to degrading or dehumanizing sexual treatment results in harm, particularly to women and therefore to society as a whole” (p. 479). In this regard, at least, the reasonable apprehension of harm test was unnecessary. The difficulty which required the application of the reasonable apprehension of harm test was determining exactly what representations of a sexual nature were degrading or dehumanizing. Sopinka J. quotes, at p. 481, from Wilson J.’s judgment in Towne Cinema Theatres Ltd. v. The Queen, 1985 75 (SCC), [1985] 1 S.C.R. 494:

The problem is that we know so little of the consequences we are seeking to avoid. Do obscene movies spawn immoral conduct? Do they degrade women? Do they promote violence? The most that can be said, I think, is that the public has concluded that exposure to material which degrades the human dimensions of life to a subhuman or merely physical dimension and thereby contributes to a process of moral desensitization must be harmful in some way. It must therefore be controlled when it gets out of hand, when it becomes “undue”.

Sopinka J. devised a test, at p. 485, which stated that depictions of a sexual nature would not be protected by the Charter where society perceived such depictions as likely to “predispos[e] persons to act in an anti-social manner as, for example, the physical or mental mistreatment of women by men”. He found the anti-pornography provision in that case to be rationally connected to the valid objective of the legislation because Parliament had a “reasoned apprehension of harm” (p. 504) that sexual representations of women which were degrading affected men’s attitudes in such a way that encouraged degrading treatment of women.

116 In the context of that decision, there was substantial evidence that depiction of degrading treatment of women leads, to an indeterminate extent, to degrading treatment of women in society. Although the precise nature of the link between depiction and attitudes, and from attitudes to actual harmful behaviour towards women, could not be conclusively proven, there was evidence that this harm actually occurred. The presumption also accords with certain logical inferences and shared perceptions of human behaviour which we might simply call “common sense”. Sopinka J. concluded that the line between permissibly and impermissibly degrading representations of sexuality should be based on a collective social understanding of what Canadians believe could lead to anti-social behaviour. While courts should not use common sense as a cover for unfounded or controversial assumptions, it may be appropriately employed in judicial reasoning where the possibility of harm is within the everyday knowledge and experience of Canadians, or where factual determination and value judgments overlap. Canadians presume that expressions which degrade individuals based on their gender, ethnicity, or other personal factors may lead to harm being visited upon them because this is within most people’s everyday experience. In part, this is because of what we know and perhaps have experienced in our own lives about degrading representations of our personal identity. In part, it is because we know that groups which have historically been disadvantaged in economic or social terms are vulnerable to such expression. In part, it is because our values encourage us to be solicitous of vulnerable groups and to err on the side of caution where their welfare is at stake. In part, it is based on the short logical leap that degrading representations, and exhortation of certain views which degrade the humanity of others, can beget that behaviour. It is also because we know that such representations and exhortations can themselves be harmful for those who are forced to endure the heightened risk of harm. As McLachlin J. put it in Keegstra, supra, at pp. 857-58:

To view hate propaganda as “victimless” in the absence of any proof that it moved its listeners to hatred is to discount the wrenching impact that it may have on members of the target group themselves. For Jews, many of whom have personally been touched by the terrible consequences of the degeneration of a seemingly civilized society into unparalleled barbarism, statements such as Keegstra’s may raise very real fears of history repeating itself. Moreover, it is simply not possible to assess with any precision the effects that expression of a particular message will have on all those who are ultimately exposed to it. . . . These considerations undermine the notion that we can draw a bright line between provisions which are justifiable because they require proof that hatred actually resulted, and provisions which are unjustifiable because they require only an intent to promote hatred.

Common sense reflects common understandings. In these cases dealing with pornography and hate speech, common understandings were accepted by the Court because they are widely accepted by Canadians as facts, and because they are integrally related to our values, which are the bedrock of any s. 1 justification. As a result, the Court did not demand a scientific demonstration or the submission of definitive social science evidence to establish that the line drawn by Parliament was perfectly drawn.

117 In my view, the case at bar does not approach this category of reasoning. The Canadian voter is not a historically vulnerable or disadvantaged group. Nor, as has been explained above, is the autonomy or dignity of any single group under attack from, or even facing the contrary interests of, another potentially more powerful group. Nor can it be said that there is a shared understanding amongst Canadians that a single inaccurate poll will mislead Canadians to an extent which, in the words of Wilson J. in Towne Cinema Theatres, is “undue”. I am, therefore, unable to accept that the harm which the government is seeking to prevent affects a large number of voters, or that such possible distortions are significant to the conduct of an election, without more specific and conclusive evidence to that effect. Although there is a pressing and substantial objective in this case because there is a clear possibility that some voters might be misled by such polls, I am not willing to go further and accept that the harm in this case warrants a significant level of deference to the government in fashioning means which trespass on the freedom of expression.

118 In determining whether a restriction is justified under s. 1 as minimally impairing the freedom of expression, this Court has stated (in RJR-MacDonald, supra, at para. 160):

As the second step in the proportionality analysis, the government must show that the measures at issue impair the right of free expression as little as reasonably possible in order to achieve the legislative objective. The impairment must be “minimal”, that is, the law must be carefully tailored so that rights are impaired no more than necessary. The tailoring process seldom admits of perfection and the courts must accord some leeway to the legislator. If the law falls within a range of reasonable alternatives, the courts will not find it overbroad merely because they can conceive of an alternative which might better tailor objective to infringement. . . . On the other hand, if the government fails to explain why a significantly less intrusive and equally effective measure was not chosen, the law may fail. [Emphasis added.]

The application of these words is largely informed by the contextual factors which I have already canvassed. In my view, little deference should be shown in this case where the contextual factors mentioned above indicate that the government has not established that the harm which it is seeking to prevent is widespread or significant.

119 The provision in this case is also overbroad and underbroad in relation to the purpose of the legislation. The ban imposed in this case is overbroad because it prohibits in the final three days of an election campaign the publication and use by voters of all those polls which would meet the usual standards of accuracy. Its underbreadth has already been mentioned in the rational connection analysis: the blackout period may not adequately disabuse voters of an erroneous impression left by a poll which did not disclose its methodology to critics or the public. Indeed, as a matter of logic, the utility of the ban as a period of response and criticism is gravely undermined by the failure to require the publication of methodological information. Both the Lortie Commission, supra, at p. 464, and Professor Lachapelle, supra, at pp. 154-55, recommended that methodological information be disclosed in addition to a blackout period. In his affidavit before this Court (Case on Appeal, at pp. 92 et seq.), Professor Lachapelle, at para. 29, states:

Even if sufficient information is available to assess the reliability of the survey, or if the survey results are inaccurately reported, an adequate period of time is required to effectively challenge or correct the published report. If an accurate public opinion survey is released on the day before an election, there will be no meaningful opportunity for public debate or response.

This evidence supports the view that the mandatory disclosure of methodological information combined with a blackout period fulfils the government’s purpose more effectively than the mandatory disclosure alone. What we are dealing with here, however, is a blackout period without a mandatory disclosure of methodology. In assessing whether this provision is narrowly tailored, the obvious alternative which Parliament could have adopted was a mandatory disclosure of methodological information without a publication ban. Indeed, British Columbia has enacted just such a measure, as has the State of New York in the United States (Election Act, R.S.B.C. 1996, c. 106, s. 235; NY statute cited in Lachapelle Study, supra, at p. 51). Although such a provision would still leave the door open to inaccurate poll results published immediately prior to the election having some impact, that possibility would be significantly reduced both by virtue of the reader’s initial access to those methodological data, and by the opportunity for rapid response by parties whose interests are prejudiced by the inaccurate poll. The government has not explained, however, how or whether this danger is any less than that of a poll published prior to the three-day blackout period without methodological data which is effectively immune from the reasoned criticism which the blackout period purports to allow. The failure to address or explain the reason for not adopting a significantly less intrusive measure which appears as effective as that actually adopted weighs heavily against the justifiability of this provision.

120 The respondent countered this concern by suggesting that a law requiring pollsters to publish methodological information along with the poll results is actually more intrusive on the freedom of expression than the three-day ban. I would reject this argument. Without making any ruling on the constitutionality of such a measure, I would simply refer to the words of McLachlin J. in RJR-MacDonald, supra, at para. 163: “As this Court has observed before, it will be more difficult to justify a complete ban on a form of expression than a partial ban: Ramsden v. Peterborough (City), [1993 60 (SCC), [1993] 2 S.C.R. 1084], at pp. 1105-6; Ford v. Quebec (Attorney General), [1988 19 (SCC), [1988] 2 S.C.R. 712], at pp. 772-73.” Whether a ban over a three-day period can properly be described as a “complete ban” is a subtle point. It is clear, however, that a provision which prohibited the publication of opinion polls without methodological information would be less intrusive to freedom of expression than a ban on publication of polling information during a crucial period. In the first case, the speaker has a choice: by complying with the prescribed conditions, he or she may engage in the speech. In the case of the ban, the speaker has no such choice: the information may not be expressed regardless of any choice the speaker makes. I emphasize in saying this, that I do not here pronounce on the constitutionality of such a provision. I am merely rejecting the respondent’s claim that the current legislative provision is the least intrusive measure which the government could have chosen to achieve its purpose.

121 The respondent pointed to the presence of similar blackout periods on public opinion surveys in other democratic countries to support its argument that this measure fell within the permissible range of alternatives. I do not find this evidence to be highly persuasive. Although a number of countries do have such provisions, most democratic countries have minimal or no restrictions on polling information; Lachapelle Study, supra, at pp. 52-62. This may be contrasted with the evidence before the Court in Butler, supra, at p. 497, to the effect that most free and democratic countries had legislation similar to that under scrutiny. Moreover, some of the bans on publication of polling information in other countries extend for very long periods, even the entire election campaign. This suggests that the purpose of those bans may be something other than ensuring that polls are as accurate as possible, which is the only permissible objective under our Charter. Where the approach in other countries is variable, or is in some relevant way different from the legislation under scrutiny in Canada, then those legislative measures must be examined more closely to determine their precise purpose and whether those purposes are of persuasive force here. Not only may the social context be quite different from that in Canada, but also the legal context within which measures restricting the freedom of speech are evaluated may be dissimilar. In the absence of some consensus in the international context, or of evidence explaining why the provisions adopted in some other free and democratic countries are compelling given the situation in Canada, the experience of some other countries as a justification under s. 1 should not be accorded great weight. This is no more than to say that the example of those countries which do not have such provisions is of as much weight in evaluating whether the legislation is justified as those which do. The key question, once the divergence of approach in the international community is established, is whether the values of Canadian society — of which the Charter itself forms a part — are more in accord with one approach rather than the other. The respondent has not taken this extra step in his analysis of those countries which do have publication bans, and therefore, I find this evidence neutral to the outcome of this case.

122 In summary, I find that the blackout period does not minimally impair the freedom of expression guaranteed in the Charter, and is therefore not justified under s. 1. The harm which the legislature is seeking to prevent does not warrant a high degree of deference to the legislature. Unlike the situation in Butler, Ross, Keegstra, and Irwin Toy, the government is not dealing with a vulnerable group which is in danger of manipulation or abuse because of an essential opposition of interests, or because of the nature of the speech itself. There were other measures which would have achieved the government’s purpose equally well or even better than the publication ban, and which would have been far less intrusive to the freedom of expression. Finally, the experience of the international community is inconclusive.

6. Proportionality Between the Deleterious Effects and the Benefits of the Ban

123 The third stage of the proportionality analysis was originally formulated in Oakes, supra, at p. 140, as ensuring a general proportionality between the measures and the pressing and substantial objective of the provision under scrutiny:

Even if an objective is of sufficient importance, and the first two elements of the proportionality test are satisfied, it is still possible that, because of the severity of the deleterious effects of a measure on individuals or groups, the measure will not be justified by the purposes it is intended to serve. The more severe the deleterious effects of a measure, the more important the objective must be if the measure is to be reasonable and demonstrably justified in a free and democratic society.

This formulation has been criticized as merely duplicating what is already accomplished by the first two stages of the proportionality analysis. As a practical matter, this is confirmed by the jurisprudence of this Court: there appears to be no case in which a measure was justified by the first two steps of the proportionality analysis, but then found unjustified by an application of the third step.

124 More recent cases have reformulated the third stage of the proportionality analysis in order to give it a distinct scope and function. In Dagenais, supra, at pp. 887-88, Lamer C.J. articulated the test as follows:

. . . I believe that the third step of the second branch of the Oakes test requires both that the underlying objective of a measure and the salutary effects that actually result from its implementation be proportional to the deleterious effects the measure has on fundamental rights and freedoms. A legislative objective may be pressing and substantial, the means chosen may be rationally connected to that objective, and less rights-impairing alternatives may not be available. Nonetheless, even if the importance of the objective itself (when viewed in the abstract) outweighs the deleterious effects on protected rights, it is still possible that the actual salutary effects of the legislation will not be sufficient to justify these negative effects. [Emphasis in original.]

He went on, at p. 889, to state:

I would, therefore, rephrase the third part of the Oakes test as follows: there must be a proportionality between the deleterious effects of the measures which are responsible for limiting the rights or freedoms in question and the objective, and there must be a proportionality between the deleterious and the salutary effects of the measures. [Emphasis in original.]

In my view, the first part of this reformulation is already achieved by virtue of the first two parts of the Oakes proportionality test. The subsequent development of the Oakes test, particularly the broad contextual approach which has been adopted by this Court since the decision in the Edmonton Journal case, ensures that the rational connection and the minimal impairment tests are sufficient to determine whether there is a proportionality between the deleterious effects of a measure, and its objective. Once a determination had been made that there is a pressing and substantial objective behind the infringing measure which may justify some infringement of the Charter, then the first and second stages of the Oakes proportionality analysis assess whether there is a coherence and an efficiency between those measures and the justified purpose. The relevant efficiency measured at the second stage of the proportionality analysis is whether it infringes on the Charter right to the minimum extent possible while still fulfilling the justified purpose.

125 The third stage of the proportionality analysis performs a fundamentally distinct role. Determining whether there is a pressing and substantial objective behind the provision under scrutiny necessarily occurs in the abstract, before the specific nature of the legislation and its impact on the Charter right has been analysed. Of course, ascertaining that objective requires a consideration of what the provision actually does, as well as documentary evidence as to what the legislator thought it was doing. Moreover, the relevant purpose is the purpose specific to the provision which limits the Charter right. But the purpose must, nevertheless, be articulated abstractly because a purpose is a goal or outcome which, by definition, may be achieved in different ways. Before the specific effects of the measure in question have been scrutinized and concretized through the first two steps of the proportionality analysis, it is often difficult to assess, in the abstract, the possible impact on Charter freedoms of a laudable legislative objective. The focus of the first and second steps of the proportionality analysis is not the relationship between the measures and the Charter right in question, but rather the relationship between the ends of the legislation and the means employed. Although the minimal impairment stage of the proportionality test necessarily takes into account the extent to which a Charter value is infringed, the ultimate standard is whether the Charter right is impaired as little as possible given the validity of the legislative purpose. The third stage of the proportionality analysis provides an opportunity to assess, in light of the practical and contextual details which are elucidated in the first and second stages, whether the benefits which accrue from the limitation are proportional to its deleterious effects as measured by the values underlying the Charter. As Professor Jamie Cameron states (“The Past, Present, and Future of Expressive Freedom Under the Charter” (1997), 35 Osgoode Hall L.J. 1, at p. 66):

. . . this branch of the section 1 analysis asks an important question. By assessing the proportionality of its deleterious effects and salutary benefits it considers, in direct and explicit terms, whether the consequences of the violation are too great when measured against the benefits that may be achieved. As such, it is the only part of the current analysis to acknowledge the harm or cost of justifiable limits: that a constitutional right has been violated.

The deleterious effects which may arise from an infringement of the Charter may be general, in the sense that any serious infringement of the right to freedom of expression may impair the climate of free exchange of ideas which is an essential value of our society; or the deleterious effects may be specific, in that a particular benefit which would accrue from the speech in question is prevented. Although both of these possible effects will and should be considered in deciding whether there is a pressing and substantial objective of the legislation, the focus at that stage is more in determining whether there is a significant harm which the government is addressing. Comparing the harm which may be prevented with the harm of the infringement itself is a balancing which can most effectively take place within the context of the proportionality analysis.

126 This approach accords with the analysis in R. v. Laba, 1994 41 (SCC), [1994] 3 S.C.R. 965, where the reformulated proportionality test proposed by Lamer C.J. in Dagenais, supra, was applied. The issue in Laba was the constitutionality of a provision which made it an offence to sell stolen ore unless the defendant could establish that he or she was the lawful owner. This was challenged under s. 11(d) of the Charter. Sopinka J. writing for the Court, at p. 1006, found that the specific objective of the provision was “to facilitate the prosecution of offenders given the special problem of proof” regarding the ownership of ore. He also found that the measure was not minimally impairing of s. 11(d) even given the validity of the purpose of the legislation. But he went on to state, at p. 1011:

Even if I were persuaded that the imposition of a legal burden was clearly more effective in achieving Parliament’s objective, I would find that it fails the proportionality test because of the excessive invasion of the presumption of innocence having regard to the degree of advancement of Parliament’s purpose. [Emphasis added.]

This analysis directly addresses the relationship between otherwise justified measures and the extent of harm to the Charter right. This weighing exercise necessarily admits of some subjectivity, but this is lessened by the analysis of the purposes, the rationality, and the efficiency of the legislation required by the previous stages of the Oakes test.

127 The impact on freedom of expression in this case is profound. This is a complete ban on political information at a crucial time in the electoral process. The ban interferes with the rights of voters who want access to the most timely polling information available, and with the rights of the media and pollsters who want to provide it. It is an interference with the flow of information pertaining to the most important democratic duty which most Canadians will undertake in their lives: their choice as to who will govern them. Such a polling ban also sends the message that the media in their role as a reporter of information, and not as an advertiser, can be muzzled by the government. Rather than approaching the problem of inaccurate polls as a question of too little information, or added incentives for preventing the publication of inaccurate polls, the government constrains the range of evaluations that a voter is permitted to make in fulfilling their sacred democratic function as a citizen. It justifies such a measure on the basis that some indeterminate number of voters might be unable to spot an inaccurate poll result and might rely to a significant degree on the error, thus perverting their electoral choice.

128 In my analysis above, I have found that despite the many factors militating against the realization of this misinformed choice, including the presence of many other polls in the public domain both prior to and concurrent with the erroneous poll, such a scenario is conceivable for a small number of voters. However, in my view, the government cannot take the most uninformed and naïve voter as the standard by which constitutionality is assessed. To quote the words of Frankfurter J. speaking in another context: “The incidence of this enactment is to reduce the adult population of Michigan to reading only what is fit for children” (Butler v. Michigan, 352 U.S. 380 (1957), at p. 383). Just as Frankfurter J. could not accept that the standard of indecency for adults should be determined according to the vulnerability of a child, nor can I accept here a measure which decides that information which is desired and can be rationally and properly assessed by the vast majority of the voting electorate should be withheld because of a concern that a very few voters might be so confounded that they would cast their vote for a candidate whom they would not have otherwise preferred. That is to reduce the entire Canadian public to the level of the most unobservant and naïve among us. This concern is also very remote from any danger that the guarantee of effective representation will be undermined.

129 My view is that, given the state of the evidence before the Court on this issue, the postulated harm will seldom occur. The benefits of the ban are, therefore, marginal. The deleterious effects are substantial. First, the ban sends the general message that the media can be constrained by government not to publish factual information. Second, it interferes with the reporting function of the media with respect to the election, which is an interference with the freedom of the media when its participation is most crucial to self-governance. These are the deleterious effects as they relate to the freedom of the speaker. But third, the ban denies access to electoral information which some voters may consider very useful in deciding their vote. If they feel that their votes are better informed as a result of having this information, then the ban not only interferes with their freedom of expression, but with their perception of the freeness and validity of their vote. This undermines the very faith in the electoral process which the government suggests is one of the rationales for this ban.

130 In my view, the doubtful benefits of this ban are outweighed by its significant and tangible deleterious effects and therefore is not justified under the third stage of the proportionality analysis. The very serious invasion of the freedom of expression of all Canadians is not outweighed by the speculative and marginal benefits postulated by the government. This is not to say that there is no possibility for Parliament to legislate with regard to the dangers that represent bad polls. As mentioned earlier, the present legislation was found to be defective not with regard to its purpose, but with regard to the fact that the means chosen to carry out that purpose did not satisfy the minimal impairment and proportionality tests.

VII. Disposition

131 I find that s. 322.1 of the Canada Elections Act infringes s. 2(b) of the Charter and that it is not a reasonable limit on freedom of expression under s. 1. Accordingly, I would allow the appeal with costs and declare s. 322.1 of the Canada Elections Act to be inconsistent with the Charter and hence of no force or effect by reason of s. 52 of the Constitution Act, 1982. I would answer the constitutional questions in the following manner:

1. Does s. 322.1 of the Canada Elections Act, R.S.C., 1985, c. E-2, as amended, infringe s. 2(b) and/or s. 3 of the Canadian Charter of Rights and Freedoms?

Answer: Yes with respect to s. 2(b), and no comment with respect to s. 3.

2. If s. 322.1 of the Canada Elections Act infringes s. 2(b) and/or s. 3 of the Canadian Charter of Rights and Freedoms, is s. 322.1 a reasonable limit prescribed by law as can be demonstrably justified in a free and democratic society for the purposes of s. 1 of the Canadian Charter of Rights and Freedoms?

Answer: No.

Appeal allowed with costs, Lamer C.J. and L’Heureux-Dubé and Gonthier JJ. dissenting.

Solicitors for the appellants: McCarthy Tétrault, Toronto.

Solicitor for the respondent: The Attorney General of Canada, Toronto.

Solicitors for the intervener the Attorney General of British Columbia: Arvay Finlay, Victoria.

Solicitor for the intervener the Canadian Civil Liberties Association: Sydney L. Goldenberg, Toronto.

• Sopinka J. took no part in the judgment.