Beyond the Ballot Box: Sexual Harassment and Legislative Accountability in Canadian Politics

Abstract

Since the #MeToo movement, several countries have taken steps to address sexual harassment in politics (for example, the United States, United Kingdom and Canada). While researchers have evaluated the electability of candidates accused of sexual harassment, less is known about what the public thinks should happen when elected officials engage in this behaviour. Utilizing an innovative module from the 2019 Canadian Election Study, we assess the steps voters believe legislatures should take when an MP sexually harasses someone. Our results demonstrate that a vast majority of the public believes that MPs should face consequences when they commit sexual harassment, including potential removal from office. We also find that women are more likely than men to believe MPs should be punished when they are accused of sexual harassment. These findings have relevance for legislatures globally, revealing the importance of transparent, independent processes to address harassment and violence in the political sphere.

Résumé

Depuis le mouvement #MeToo, plusieurs pays ont pris des mesures pour lutter contre le harcèlement sexuel en politique (par ex., les États-Unis, le Royaume-Uni et le Canada). Alors que les chercheurs ont évalué l'électabilité des candidats accusés de harcèlement sexuel, on sait moins ce que le public pense qu'il devrait se passer lorsque les élus adoptent ce comportement. À l'aide d'un module innovant de l'Étude sur l’élection canadienne 2019, nous évaluons les mesures que les électeurs pensent que les assemblées législatives devraient adopter lorsqu'un député harcèle sexuellement quelqu'un. Nos résultats démontrent qu'une grande majorité du public pense que les députés devraient subir des conséquences lorsqu'ils commettent des actes de harcèlement sexuel, y compris une éventuelle révocation. Nous constatons également que les femmes sont plus enclines que les hommes à penser que les députés devraient être punis lorsqu'ils sont accusés de harcèlement sexuel. Ces résultats sont pertinents pour les législatures du monde entier, car ils révèlent l'importance de processus transparents et indépendants pour lutter contre le harcèlement et la violence dans la sphère politique.

Introduction

Over the last several years, sexual misconduct allegations against politicians have surfaced in many countries, spurred on by the #MeToo movement.¹ In Canada, in 2020 a sitting Liberal MP was arrested on assault, break and enter, and criminal harassment charges. After the charges were made public by the media, the MP announced he would be “stepping back from the Liberal caucus” but not resigning as an elected official (Humphreys, 2020). It later emerged through an internal party investigation that he had previously sexually harassed a woman staff member, yet the party still approved him to seek re-election in 2019 (Burke, 2021). In another case—and just days before the 2021 federal election—the media reported that a candidate had previously faced sexual assault allegations, yet he too had received a green light from his party to run in the election. He now serves as an Independent MP.² These examples are preceded by other instances of politicians who have engaged in sexual misconduct and include representatives from all the major parties and every level of office in Canada (see Collier and Raney, 2018a, 2018b).

In this article, we examine public opinions on how legislatures should respond when an elected official engages in one form of sexual misconduct: sexual harassment (SH).³ US-based research offers important insights into the electability of candidates who commit SH, with some voters willing to punish sexual harassers at the ballot box (Stark and Collignon, 2022; Holman and Kalmoe, 2021). Less well known are the public's expectations of how SH should be addressed by legislatures after a candidate has been elected. We ask two questions: (1) What does the public believe should happen when an elected official engages in sexual harassment between elections? (2) Do women and men share the same expectations about how sexual harassment in politics should be handled?

Heightened media focus on the #MeToo and #TimesUp movements has generated considerable public attention on these issues globally. In the #MeToo era, it is plausible to assume that members of the public expect legislatures to take SH seriously when it occurs in the political realm. Opinions on SH may differ between women and men, especially at the time an allegation is made. Since women are more likely to be victims of SH compared to men, their life experiences and gendered positionalities can inform their perceptions of workplace SH (Masuoka et al., 2021). Women are often reluctant to report being sexually harassed for many reasons, including a fear of not being believed, fear of retaliation, perceived unfair and biased processes, and a belief that the perpetrator may not face serious consequences for their actions (Bergman et al., 2002; Butler, 2008). These sentiments are captured by the #BelieveWomen and #BelieveSurvivors hashtags, which emerged in 2018 and were intended to debunk the myth that women falsely report sexual misconduct and should be believed at the time an allegation is made (Bolinger, 2021).⁴ We therefore might expect women to believe a SH allegation at face value and to have higher accountability expectations in how legislatures handle such claims when compared to men.

To examine our research expectations, we employ survey data and analyze attitudinal differences in legislative responses to SH at two stages of a claim: first, when an MP has been accused of SH, and second, when an MP has been found through an independent investigation to have committed SH. The first scenario allows us to assess what the public believes should happen when an MP has been accused of SH. When controlling for other socio-demographic and political variables, we find that women are more likely than men to expect an elected official to face consequences at the time aharassment allegation is made. Compared to men, women express greater support for public disclosure of a SH allegation and for an MP's suspension pending an investigation. This finding aligns with other gaps observed between women's and men's political behaviours and attitudes, including gaps relating to voter turnout, partisanship, political ambition and political interest (Studlar et al., 1998; Inglehart and Norris, 2003; Fox and Lawless, 2014; Bos et al., 2022). The documentation of an accountability gap between women's and men's attitudes on how legislatures should handle SH allegations against an elected official is a novel contribution to gender and politics scholarship.

The second scenario assesses what voters believe should happen when an MP is found through an independent investigation to have engaged in SH. By focusing on two different stages of a SH claim, our study offers new evidence of public attitudes about the process of how SH claims against elected officials should be handled by lawmakers, as well as the expected outcomes of such cases. Interestingly, in this second scenario, the gap between women and men we observe in the first, or “alleged,” scenario disappears. After an independent investigation SH finding, we find that the vast majority of voters, women and men, support legislative action to address the issue, including an MP being removed from office to face a re-election by their constituents. Taken together, these results confirm our argument that when an elected official engages in SH, both women and men expect them to be held accountable and that women hold higher expectations of accountability than men at the time an allegation is made.

Understanding public opinion on SH in politics has relevance for democratic representation and legislative accountability. Legislatures that tolerate SH could be perceived by some members of the public as hostile spaces, especially for those who are more frequently targeted by violent and harassing behaviour. Today, white, heterosexual, cis-gendered men are overrepresented in the Canadian House of Commons, with just 103 women MPs (30.5%), twelve Indigenous MPs (4%), eight Black MPs (2%) and eight MPs who identify as queer, trans or two-spirit (2%).⁵ The perception that politicians are turning a blind eye to SH within their ranks could undermine efforts to create more inclusive, diversity-sensitive parliaments and to build and maintain public trust in Canada's legislatures more broadly.

Below we situate our research in the relevant literatures, introduce our data and present our analyses. We conclude by offering some best practices on how lawmakers can achieve legislative accountability on this issue. Recommendations include transparent, independent processes and the need for real consequences for elected officials who engaged in sexually harassing behaviour.

Public Opinion on Sexual Harassment in the #MeToo Era

To date, gender and politics research focuses mainly on how and whether voters punish candidates who engage in SH at the ballot box. Studies point to a mix of partisanship and gender as important predictors of voter behaviour in this regard. In the United States, Holman and Kalmoe (2021) find that voters who viewed the “other party” as better at addressing this problem were more likely to abstain or defect to the other party. In another US-based study, Cossette and Craig (2020: 122) observe that harassment allegations contributed to a loss of voter support for both women and men politicians equally. However, Democrats were more likely than Republicans to express anger or sadness that harassment still occurs and to suggest that victims need to be taken at their word. They also find that women voters held candidates accused of harassment to a higher standard than did men (111). Stark and Collignon (2022) find that Republican voters were more likely than Democratic voters to support a candidate from their own party irrespective of such allegations. Masuoka et al. (2021) similarly find that Democratic women were more likely than other voters to punish candidates who had been accused of SH.⁶

Public opinion research on political scandals (and “sex” scandals specifically) similarly finds differences between women's and men's attitudes. Several studies indicate that women appear to be less tolerant of political corruption and more likely to support penalties for corrupt politicians and parties than are men (Allen and Birch, 2012; Stensöta et al., 2015; Alexander et al., 2020). Women are also less likely than men to vote for a candidate who has been involved in some type of scandal (Barnes et al., 2020, 112). In addition to varying based on partisanship, voters’ willingness to penalize politicians who engage in sexual misconduct varies based on age, political ideology and attitudes about sexism (Costa et al., 2020).

Zeroing in on SH in politics specifically, Cosette and Craig (2020) find that when an elected official has been found to have engaged in SH, voters will support their resignation or removal from office. Looking at a broader set of sanctions for similar cases, a 2019 Fawcett Society report found that 7 in 10 British respondents believed that an MP should be required to issue an apology, that a by-election should be triggered, and that the person should be removed from office and banned from running again for a period of time (Culhane, 2019). Using this broader range of sanctions, the present study reports, for the first time, findings from multivariate analyses that evaluates what the public believes should happen when there has been an allegation of SH against an elected official or when an official has been found to have engaged in SH in Canada.

We further draw on workplace SH literature to formulate our research expectations, which finds differences between women's and men's perceptions of this behaviour. Lonsway et al.'s (2008: 612) research shows that men are more likely than women to accept various myths about SH, including the belief that women who report being sexually harassed are exaggerating the behaviour or lying. Similarly, Fitzgerald and Shullman (1993) suggest that attitudinal differences between women and men about SH are likely associated with gendered societal norms and expectations, whereby sex-based harassment is linked to gendered structures of power, discrimination and privilege. Gender socialization is thought to be a key determinant of sexual harassment, whereby: “men assert power and dominance over women both at work and in society” (Welsh, 1999: 176). Women are also more likely than men to describe a wider range of behaviours as SH and are generally less tolerant of such behaviour (Fitzgerald and Shullman, 1993; Bitton and Shaul, 2013).

Public opinion data on #MeToo corroborate the importance of attitudinal differences between women and men in relation to workplace SH. Statistics Canada's “2020 Survey on Sexual Misconduct at Work” found that women were more likely to experience workplace harassment than men. The study also found that women were more likely than men to express anxiety about not being believed and to fear retaliation, embarrassment and shame, as well as negative career implications; these were all reasons not to report their experiences (Statistics Canada, 2021). Compared to women, men have been found to be more likely to say that new rules on SH are “killing the human element at work” and that “people's careers are being ruined without due process and a chance to defend themselves” (Angus Reid Institute, 2018). A 2019 poll found that one of the biggest workplace concerns men have is being falsely accused by a woman colleague of harassment or assault (Elsesser, 2019).⁷ Men are also more likely than women to be suspicious of SH claims (Kam and Archer, 2021). In their study, Stark and Collignon (2022: 344) find that women were more likely than men to describe political candidates accused of sexual assault/harassment using words that signal their belief in the allegation, referring to them as “shady,” “untrustworthy,” a “predator” and a “creep.” Given this research, we postulate that women will be more likely than men to believe an allegation of SH and to expect a swift legislative response to deal with the accusation accordingly.

The present study contributes to existing scholarship in several ways. First, building on the work of Cossette and Craig (2020), our survey asks respondents their opinions about the punishments legislatures can potentially impose upon their members, rather than about whether they might punish a candidate by voting for someone else on election day. In other words, our article focuses on legislative, rather than electoral, accountability in addressing SH in politics. Second, we consider attitudes about legislative remedies to address SH following both an allegation and a finding of harassment through a formal investigation—a distinction that, to our knowledge, has not received much academic attention. These potential remedies include a broader array of sanctions than have previously been assessed, including temporary suspension, public disclosure of allegations, demanding an apology, and removing an MP from office in order to face a by-election.⁸

Data and Results

To assess public expectations of legislative accountability on SH, we ask two questions: (1) What does the public believe should happen when an elected official engages in sexual harassment between elections? (2) Do women and men share the same expectations about how sexual harassment in politics should be handled? To answer these questions, a unique module was designed and implemented as part of the 2019 Canadian Election Study (CES). The sample contains 1,146 respondents who were eligible voters. Data were collected by way of an online survey fielded during the campaign period, from September 18 to October 20, 2019 (Stephenson et al., 2020).⁹

The module contains a series of four questions previously not asked in Canada and which were adapted from a study of attitudes on SH in British politics discussed above (Culhane 2019). Respondents were first asked the following two agree/disagree questions about a hypothetical scenario where their MP was accused of SH:

If your Member of Parliament (MP) was accused of sexual harassment what action do you think should be taken?

1. 1) The public should be made aware that a complaint has been made against the MP, while maintaining the privacy of the complainant.

2. 2) The MP should be temporarily suspended from their work duties until an investigation has concluded.

Next, respondents were asked two more agree/disagree questions, this time about potential consequences if their MP were found, through an investigation, to have sexually harassed someone:

If your Member of Parliament (MP) was found to have sexually harassed someone after an independent investigation, what action do you think should be taken?

1. 1) The MP should be required to issue a public apology.

2. 2) The MP's constituents should be able to trigger an election in their constituency, so that local people can decide if they want a new representative.

These four survey questions form the basis of our analysis and allow us to investigate the level of support for each consequence, as well as the correlates of this support. Our analysis proceeds in multiple stages. First, we perform a descriptive analysis to determine levels of support by respondents’ sex across our four accountability variables. Next, we consider the correlates of these attitudes in a multivariate analysis. Here, we include a variety of other socio-demographic and attitudinal controls in our models. As part of this analysis, we consider differences in the correlates of support for various accountability measures following (a) accusations and (b) findings of SH. We provide more detail on these analyses below. Appendix I contains a full list and description of all survey questions used. Results are weighted for region, gender, and age.

Descriptive analysis

Prior to considering the correlates of support for the accountability of MPs, we describe aggregate level attitudes toward each measure. Table 1 shows the frequency distributions for support for the responses to accusations (“public awareness of accusations” and “suspension”) and findings of SH (“public apology” and “force election”). For each question, we show support for all five response categories, and then we pool directional responses—that is, strongly (dis)agree and somewhat (dis)agree frequencies are combined. Regardless of the differences between women and men that may exist on these measures, this descriptive information provides important context into the level of support for various consequences that MPs accused of, or found to have committed, SH should face.

Table 1. Attitudes toward MP Accountability following Accusations and Findings of Sexual Harassment

Note: N = 758. Not all columns add to zero, due to rounding.

Results in Table 1 show that, at the aggregate, there is considerable public support for accountability of politicians involved in SH scandals, at both stages of a claim. The modal response for all our variables is “strongly agree.” Across all four variables, more than three-quarters of respondents agree with the proposed consequence, and very few disagree. At the aggregate level, there is no discernible difference between levels of support for the “accusation” consequences, as compared to those that apply to MPs who are “found” to have committed SH (though this is not necessarily surprising, since the consequences following such a finding are somewhat more severe than for accusations). The substantiated finding here is, therefore, that there is widespread support for these accountability measures when an MP is involved in a SH investigation.¹⁰

Socio-demographic analysis

Having established that there is a low level of tolerance for sexually inappropriate behaviour among MPs, we next consider if and how there are differences between women and men respondents related to the two sets of consequences variables. We answer these questions by way of a series of ordered logistic regression models, with results shown in Table 2. We run two models for each of the four agree/disagree accountability variables. The first is a simple bivariate analysis, with a respondent's sex as the only explanatory factor. This serves to provide a baseline description of the relationship between women/men and our outcomes of interest.¹¹ Second, we conduct a multivariate analysis, where we include a variety of other socio-demographic characteristics: age, education, income, marital status, sexual orientation, employment status, immigrant status, race (dummies for racialized and Indigenous, with white as the base) and language. Models of vote choice commonly include a wide range of socio-demographic characteristics (Anderson and Stephenson, 2010; Gidengil et al., 2012), and each of these factors has been found to be related to vote choice in different contexts. There is therefore good reason to expect that these factors matter when considering other political variables, such as those under consideration here. A further benefit of considering these other socio-demographic indicators is that they allow us to place any potential differences between women's and men's responses in context. If a respondent's self-reported gender matters more or less than these other variables in our models, then this would speak to the relative importance of this variable in shaping attitudes toward punishment following claims of sexual harassment.

Table 2. The Correlates of Attitudes toward Accountability of MPs for Sexual Harassment: Ordered Logistic Regression Results

Note: N = 758 in all models. Entries show coefficients and standard errors (in parentheses).

* p < .05; ** p < .01

The multivariate models serve two functions. First, they enable us to control for any other factors that might covary with a respondent's self-identified gender, allowing us to properly isolate the relationship between this variable and our attitudes of interest. Second, they allow us to determine if any other factors might be associated with our outcome variables—to provide further context to any relationships we may observe between women and men respondents. The models in Table 2 show ordered logistic regression coefficients, but we also present the results for these models via marginal effects in Appendix II. As a test of robustness of our findings, we also ran all eight models using OLS. Readers can refer to Appendix III for confirmation that the substantive conclusions of this analysis remain unchanged with this alternative approach.

Table 2 contains several findings of note. The most striking is that a respondent's self-identified gender has a statistically significant relationship to two of the four accountability variables. Specifically, this variable is relevant for the questions about SH allegations. Women are more likely than men (both in the bivariate and multivariate models) to agree that the public should be made aware of SH allegations and that MPs should be suspended following an allegation (at the 95 per cent level in all four models). The strength of these findings is reinforced by the paucity of statistically significant results for other variables, in any of the other eight models in the table. Only one other socio-demographic characteristic (language) is significant in two models, and the direction of the effect is different in those two instances.¹² The most consistent result in Table 2, therefore, is that women are more likely than men to support holding MPs accountable following allegations of SH. In contrast to these strong findings are the null results in the “finding of SH” models. That is, women are no different than men when asked whether they agree that MPs should make an apology or if constituents should be able to hold a recall election. These findings show that after an independent investigation has concluded that an MP committed SH, the attitudinal gap between men and women disappears, with men just as likely as women to support the imposition of various sanctions against the MP in question.

Observant readers might note that, though not statistically significant, the coefficients for women and men in the “finding” models are positive (thus in the same direction as in the “accused” models). We have therefore conducted an additional test to confirm that the differences in the “accused” and “finding” models are statistically significant for this variable. Put another way, the test allows us to determine if the gap between women and men is, in fact, greater in the “accused” models. To this end, a new composite variable was created that combines responses to all four questions and taps into the difference in attitudes toward the two “finding” and the two “accused” variables (that is, values for the “accused” variables were subtracted from the “finding” responses). The composite variable was then regressed onto respondents’ sex and the same control variables considered in Table 2. Results reveal that a respondent's self-identified gender is significantly related to this new variable, providing further evidence that the gap between women and men observed in the “accused” models is significantly different than in the “finding” models (see Appendix IV for more details on this analysis).

Having established that there is a unique gap between women and men in the “accused” models, we return briefly to Table 2 to try to make sense of the magnitude of these observed differences. As ordered logistic regression coefficients are unintuitive to interpret, we present Figure 1, which shows the predicted probability of respondents falling into each agree/disagree category for the two variables for which a respondent's self-reported gender is statistically significant—the two “accused” variables. Results are calculated using post-estimation, following the full models in Table 2. In Figure 1, we vary the results between women and men, leaving the values of the other variables unchanged.

Figure 1. Predicted distribution of attitudes toward the consequences MPs should face following allegations of sexual harassment, by respondent's sex

Unsurprisingly given the results in Table 1, we see that the vast majority of individuals support all forms of accountability for MPs following a SH accusation. There are, however, noteworthy differences between women and men respondents. For both variables, women are more likely to agree with the proposed consequences than are men. The gap is an estimated 5.4 percentage points higher for women, compared to men, in terms of making the public aware of an allegation, and it is 7.0 percentage points higher for an MP's suspension. Figure 1 makes clear that the source of this difference is the “strongly agree” category. Women are considerably more likely than men to fall into this category, on both variables. Women are therefore not only more likely than men to agree with holding MPs accountable in the face of accusations of SH, they are also more likely to strongly agree with this course of action.

Attitudinal analysis

The next step in this research is to gain a deeper understanding of the relationships observed in Table 2. The explanatory variables included in the models above are socio-demographic rather than attitudinal. That is, they tell us who is comparatively likely to favour SH accountability (women) but not why. There are compositional differences between women and men on the basis of several attitudinal dimensions that might help to account for the differences observed between women and men respondents in Table 2 and Figure 1, and it is worth attempting to identify the attitudinal sources of these patterns.

A novel dataset that includes a variety of measures specific to this research agenda is required to thoroughly conduct such an examination. Existing datasets such as the CES, as invaluable as they are, are not custom made for an analysis of this nature and can only tell us so much about this phenomenon. Still, the CES does include variables that allow for some limited probing on this front, including measures of partisanship, ideology, and attitudes toward feminists. Existing research reveals that partisanship, ideology, and attitudes toward sexism are relevant in determining levels of support for candidates who have engaged in sexual misconduct, with right-leaning, conservative voters exhibiting higher levels of sexism and thus higher levels of tolerance for this behaviour (see Doherty et al., 2011; Barnes et al., 2020). In the absence of a survey question on sexism, we use “attitudes toward feminists” as a general, albeit crude, proxy. Since SH is fundamentally an abuse of gendered power, this variable is presumably able to capture some generalized sentiment about sexism. Further, by including partisanship and ideology simultaneously, we are able to test if these variables are responsible for the observed differences between women and men respondents as seen in Table 2. If they are responsible, then we should see the differences between women's and men's responses disappear.

Though not our primary focus here, we are also interested in testing to see how these variables are related to the “punishment” indicators that serve as our outcome variables. As such, we present Table 3, which shows models similar to those for the “accused” questions in Table 2, with the addition of the three attitudinal factors that we expect will vary with a respondent's self-identified gender and, potentially, attitudes toward the consequences MPs should face when accused of SH. If partisanship, ideology, and attitudes toward feminists are responsible for the attitudinal gap between women and men, then the “women” variable should cease to be statistically significant in the expanded models. Otherwise, other explanations for this gap need to be found. Note that all variables have been coded to range from 0 to 1. Full details on variable coding are found in Appendix I.

Table 3. Socio-demographic and Attitudinal Correlates of Support for Accountability When an MP Is Accused of Sexual Harassment: Ordered Logistic Regression Results

Note: N = 533. Entries show coefficients and standard errors (in parentheses).

† p < .07; *p < .05; **p < .01

Our findings reveal that partisanship, ideology, and attitudes toward feminists do not account for the differences between women and men observed here; the “women” variable in Table 3 remains statistically significant after the attitudinal variables are added. The variable is significant at 95 per cent in the “suspension” model, and while the p-value increases to just above that level (to .07) in the “public awareness” model, this result is not unexpected given the decline in N to 533 cases (the coefficient is almost unchanged, but the standard error is now higher). There is therefore little evidence that these factors explain the differences between women's and men's punishment beliefs at the allegation phase of a SH claim. As such, other explanations for the observed gap between women and men are necessary, as are new datasets to consider them. Moreover, with small exceptions—the attitudes toward feminist variable in the first model, and New Democratic Party (NDP) partisanship in the second—the attitudinal variables themselves are insignificant.¹³ These variables, which one should expect to be related to attitudes toward punishments, do not have an independent effect when considered simultaneously with the socio-demographic factors. These largely null results provide even more evidence that the differences observed here between women's and men's attitudes toward the punishment of public officeholders who engage in SH are meaningful.

Conclusion

Our study set out to answer two research questions: What does the public believe should happen when an elected official engages in SH between elections? And do women and men share the same expectations about how SH in politics should be handled? Using novel survey data from the 2019 CES, the study's findings offer several new insights into views on legislative accountability to address SH. First, we identify differences of opinion between women and men respondents on the steps they expect legislatures to take when an elected official has been accused of SH. Compared to men, women are more likely to expect legislatures to publicly disclose SH allegations and more likely to support the temporary suspension of an MP until an independent investigation is held. When controlling for other variables, identifying a gap between women's and men's attitudes on legislative accountability to address SH is a novel finding to the SH in politics literature, and it highlights the importance of fair and robust anti-harassment rules in politics, especially for women.

In light of these findings, we might ask why women, compared to men, support stronger accountability measures at the time a SH allegation is made. Although our data do not allow us to answer this question directly, we speculate that this observed difference is likely due to different lived experiences of violence and harassment. Since women are more likely than men to experience SH, they are also more likely to reject the myth that women falsely report this behaviour (for example, #BelieveWomen, #BelieveSurvivors). Anchored in this belief, women may also be more likely than men to accept prima facie an allegation of SH and to expect legislators to take precautionary measures that would protect other potential targets of SH in the future. Publicly disclosing allegations and temporarily removing a potential perpetrator (pending an investigation) would help prevent repeat offences. In contrast, (some) men's fears of being falsely accused of SH may make them more reluctant than women to believe an allegation of SH without “evidence,” thereby reducing their expectations that an MP should face consequences for their actions prior to an investigation.

We also conducted a preliminary analysis of the sources of differences between women and men attitudes on SH-related punishments. Our data indicate that partisanship, political ideology, and attitudes toward feminists do not diminish the gap between women and men observed in Table 1. This set of findings raises important questions for future research using a new dataset. While US-based studies have found partisan differences in how voters evaluate politicians who are accused of SH (see Cossette and Craig, 2020), we observe no statistically significant partisanship differences in our data, which leads us to wonder whether Canada's less polarized, multiparty context might “filter” voters’ attitudes about SH in different ways compared to the United States—and perhaps less prominently so. Future research should probe this issue. Additional research should also consider whether the sex of an accused MP affects attitudes toward legislative accountability, as well as whether the positions of the person making a complaint matter (that is, does it matter whether it is another MP, a staffer, or an old acquaintance). How do the responses of party leaders or the accused themselves affect these attitudes? In a different vein, how do the media react to such accusations and how does their coverage shape attitudes? Qualitative methods might also prove helpful in identifying the causal mechanisms behind the relationships discovered using our quantitative approach. Intersectional factors such as race and sexual orientation are also likely relevant.

Regardless of the precise reason for the gap between women's and men's attitudes when an MP has been accused of SH, the disappearance of this gap at the findings stage of a SH is noteworthy. One possible interpretation for the lack of differences between women's and men's attitudes at the findings phase of a SH claim could be that the punishments under consideration are not especially harsh. For example, allowing a politician to apologize could afford them the opportunity to frame the story on their own terms, or even to justify their behaviour. Similarly, holding a by-election could result in a politician who has committed SH being re-elected. Perceived tolerance of SH in politics may lower the public's belief (and perhaps especially women) that these punishments are sufficient to hold the MP in question accountable (see Fitzgerald and Shullman, 1993; Clarke, 2014). Future research might consider other, and potentially more serious, consequences for sexually harassing behaviour, such as those that a party can impose upon one of its elected members.¹⁴

Nonetheless, in terms of potential legislative steps to be taken, our findings point to widespread public support for legislatures to take actions following both an allegation and a finding of SH. There is little support for leniency, and the overwhelming majority of respondents favour all types of MP consequences for those who engage in SH; time is clearly up for MPs in both the alleged and findings scenarios. This finding contributes to existing research and global advocacy on violence and harassment in politics, where relatively little is known about public expectations of legislatures in the #MeToo era. Using public opinion data, our findings support global calls to action for strong measures to address gender-based violence in politics (National Democratic Institute, 2016; Inter-Parliamentary Union, 2019; Raney and Collier, 2021).

Finally, our results offer some concrete solutions to improve democratic accountability in Canada. The adoption of a credible, visible system for investigating SH claims by legislatures could help ensure that all citizens—women and men—support the punishment of elected officials who engage in this behaviour.¹⁵ Unfortunately, these benchmarks have not yet been met in Canada or in many other legislatures globally. The Canadian House of Commons’ existing grievance and sanctioning processes that (partially) address gender-based violence in politics remain relatively opaque to the public and are not fully independent of partisan influence (Collier and Raney, 2018b). Our findings underscore the need for transparent and independent claims processes where there is a real possibility of holding elected officials who engage in SH accountable for their actions. Discussions are needed on how legislatures can use their collective parliamentary privilege (for example, their disciplinary powers) to more effectively hold elected members who engage in this unethical behaviour accountable. Such discussions should be incorporated into broader and much needed debates about modernizing parliamentary ethical rules in order to align them with the public's growing expectations of transparency and democratic accountability (Stedman, 2022: 210). In the eyes of the public, independent investigations of sexual harassment claims appear to be key tools in broader efforts to improve legislative accountability in Canadian politics.