The effect of social pressure from family and friends on turnout

Abstract

Recent research about the decision to vote or abstain finds a causal effect of social networks and social pressure. Yet this literature does not examine how this social pressure is exerted and by whom. This study aims at correcting these shortcomings. Using a two-wave panel survey conducted in Canada, we distinguish between the pressure exerted by friends and the partner and between descriptive and injunctive norms. We find that most people are not subjected to strong injunctive pressure, social pressure is most prominent in the household (between partners), it is mostly descriptive, and it has a powerful effect.

Keywords

Family friends household partner social pressure turnout

Introduction

Recent research about the decision to vote emphasizes the normative and social aspects of the phenomenon. Experimental evidence has confirmed the effects of social pressure on the probability that an individual goes to the polling station on election day. Several works related to the Get-out-the-Vote research project show that people can be persuaded to vote if they are threatened with revealing their deviant behavior to their neighbors (Gerber, Green, & Larimer, 2008). Other works have also brought forth the importance of friends, acquaintances, relatives, and partners in enforcing the social norm that voting is desirable.

This literature, mostly confined in the U.S., concludes that the effect of social pressure ranges from negligible to 8 percentage points depending on the salience of the election and the ex ante propensity to vote (Rogers, Green, Ternovski, & Young, 2017). As for the mechanisms by which social pressure is exerted, most of this literature insists on the observability of deviant behavior and the threat of surveillance as conditions for social pressure to take place (Gerber, Green, & Larimer, 2010; Mann, 2010; Panagopoulos, 2014), but we still know little about how social agencies concretely promote voting at the time of an election.

This raises a series of questions regarding social networks, social pressure, and the decision to vote. The first questions are descriptive: How much social pressure are people exposed to and from whom? Second, there is the question of who is mainly responsible for successful social pressure when it comes to voting. Third, we do not know much about how the influence of social agencies takes place. In this respect, we wish to distinguish two processes, one in which pressure is explicit, that is, the person votes because her partner/family/friends tell her that she should vote (injunctive norm), and one in which pressure is implicit, that is, the person votes because people in her network vote and she follows the descriptive norm. Finally, we are interested in ascertaining how people react to that pressure, to what extent it does make individuals more prone to vote. Hence, this research addresses four questions regarding informal social pressure: how much, from whom, how, and what impact?

We address these questions by making use of a data set that includes questions about whether respondents feel subject to pressure from their social network. Our approach consists in determining whether pressure has an independent impact on the propensity to vote, controlling for the most important individual-level determinants of turnout. We wish to ascertain whether social pressure has an independent effect but also to estimate the magnitude of the effect, and to distinguish the influence of the main social agencies, while paying attention to descriptive information about the relative frequency and intensity of informal social pressure.

Theoretical framework

Most interpretations of the decision to vote or abstain treat that decision as an individual one. According to this approach, the turnout decision depends on the rational calculus of benefits and costs (Downs, 1957; Riker & Ordeshook, 1968) and the presence or absence of resources (Brady, Verba, & Schlozman, 1995). Psychological approaches add to the former the role of civic duty (Blais, 2000; Blais & Achen, forthcoming; Knack, 1995), as well as attitudes such as political efficacy, partisanship, altruism, and patience (for a review, see Smets & Van Ham, 2013), which make the individual comply with the social norm regarding electoral participation.

The existence of a social norm signaling the virtues of voting does not mean that everyone will abide by the norm. Some may need a reminder or a sanction to feel compelled to vote. This is the stepping stone of the interpretation that stresses the social aspects of turnout. This school of thought puts emphasis on the role of social pressure from family and other members of the community to make deviant individuals (potential abstainers) comply with the social norm that voting is a duty. This research strand has shown the effects of social connectedness on the likelihood of voting. Hence, weakening social ties due to aging (Goerres, 2007; Knack, 1992) or widowing (Bhatti & Hansen, 2012; Denver, 2008; Wolfinger & Wolfinger, 2008) foster abstention, as in these situations the individual is less likely to be held accountable before her community when violating the social norm regarding the virtues of voting.

From such a perspective, an individual’s decision to vote is shaped by what her social circle considers appropriate (Bond et al., 2012; Gerber et al., 2008; Nickerson, 2008; Sinclair, 2012). This research strand points at normative conditionings and the social aspects of the decision to vote, showing that social connectedness and social sanctions from family, friends, and acquaintances can make abstainers comply with the norm that the good citizen has a duty to vote (Campbell, 1982; Denver, 2008; Gerber et al., 2008; Goerres, 2007; Knack, 1995; Wolfinger & Wolfinger, 2008).

Experimental research has repeatedly demonstrated that people can be persuaded to vote in the context of a canvassing campaign. The Get-out-the-Vote (GOTV) experiments reveal that social sanctions boost turnout as long as there is interpersonal exchange (Garcia Bedolla & Michelson, 2012; Green & Gerber, 2015). Although experimental designs are ideal to identify the causal effect of a mobilizing campaign, they cannot directly manipulate social pressure exerted by one’s close social network. It is noteworthy that most of these works rely on variants of the “neighbor mailer” GOTV treatment, in which neighbors play a prominent role in the defense of the social norm (Gerber et al., 2008). Nevertheless, other works attribute more influence to friends, pointing out that being aware through social network sites that your friends have voted makes you more prone to effectively attend the polling station (Bond et al., 2012).¹ What happens in the household is also relevant; when one person in the family is mobilized to vote, the mobilizing effect is transferred in large part to the partner (Nickerson, 2008). Hence, the relative impact of different social agencies still remains unclear.

Finally, most of the experimental research on social pressure and voting is confined in the U.S., and it is difficult to tell whether the identified effects would travel to other countries, given the specificities of the American case. Furthermore, the GOTV experiments emphasize the role of disclosure (or visibility) of the deviant behavior (abstention) and the crucial impact of shame to make deviants comply (vote). This implies that individuals’ social networks are a “threatening” presence, without specifying how the social norm about voting is concretely enforced.

Surveys can help fleshing out these questions. In this regard, survey evidence has revealed that the social networks most likely to influence the individual are informal ones (Abrams, Iversen, & Soskice, 2011). According to the informal social networks model, people vote if their informal network of family and friends attach enough importance to voting, because voting leads to social approval and vice versa. Similarly, Sinclair (2012) posits that when it comes to following social norms about voting, citizens take cues from friends and family rather than from other agencies or society at large. Kenny’s (1993) work also shows that one is more prone to vote more if the people with whom she discusses politics do vote, the effect being stronger if the discussant is the partner. Other studies have shown that the household is a critical unit (Cutts & Fieldhouse, 2009; Fieldhouse & Cutts, 2012, 2016). It makes a big difference for our interpretation of informal social pressure whether that pressure is mostly confined to the household or is broader in scope. Yet, as far as we can tell, no study has systematically compared the effects of these two main agencies, friends and family.

Friends may encompass core personal friends but also acquaintances. The larger the personal network, the greater the social embeddedness and the likelihood of abiding by social norms. From Berger’s (2009) perspective, friendship—a relationship of trust and reciprocity—is a precursor of moral engagement (attention and commitment to moral principles including law abidingness), which in turn fosters turnout, among other forms of social engagement. Previous research has shown that turnout is highly correlated between friends (Berelson, Lazarsfeld, & MacPhee, 1954; Glaser, 1959; Huckfeldt & Sprague, 1995; Knack, 1992; Straits, 1990). Although the effect of this social agency depends largely on how like-minded friends are, research has shown that a single person embedded in a large social network affects the turnout decision of at least four people on average in a turnout cascade (Fowler, 2005). Friends become more important in the socialization process of the individual after childhood, taking precedence over parents until a romantic partner appears (Carbery & Buhrmester, 1998).

Then, there is the family. In the case of adults, since we are interested in those who have the right to vote, the family encompasses, in order of importance, the partner, children, parents, siblings, and at some point, more distant relatives (Buhl, 2009). The transition to adulthood (through employment, marriage, parenthood, and other responsibilities) brings a new understanding between adult children and their parents, while relations with siblings become less conflictual (Aquilino, 1997; Guan & Fuligni, 2016). Yet the lion’s share of family influence comes from the partner.

Indeed, young adults spend much more time with their partners than with their friends and parents, resulting in a greater influence of the former agency (Carbery & Buhrmester, 1998; Collins & van Dulmen, 2006). As for voting behavior, it is an established finding in the literature that there is a strong correlation between the turnout decision of spouses (Glaser, 1959; Kenny, 1993; Straits, 1990) and that living with a partner increases turnout (Milbrath & Goel, 1977; Wolfinger & Rosenstone, 1980). Recent works confirm that individuals become more alike with regard to political participation after marriage (Stoker & Jennings, 1995). Among the reasons why this happens, Pattie and Johnston (2000) point out that partners remind each other to vote and often walk to the polling station together. Yet a more moralizing mechanism may be at play, since some research points out that the moral/religious characteristics of the partner are particularly contagious (Scheepers & Van Der Slik, 1998). We can extend this to social norms about voting, which have indeed a moral component. We, therefore, expect the influence of the partner to outweigh that of friends.

Finally, there is the question of how these agencies influence the individual. Most experimental research construes social pressure as the sanctions that abstainers anticipate if they do not vote. But social pressure also encompasses the actions conducted before the election to promote the desirable (voting) behavior and that may be accompanied by no threat of sanctions. In this regard, the literature distinguishes between descriptive and injunctive norms (Cialdini, Reno, & Kallgren, 1990). While descriptive norms consist of individuals’ perceptions of how other people act, injunctive norms refer to perceptions of what people think that the individual should or should not do (Gerber & Rogers, 2009; Sheeran & Orbell, 1999). Both have been manipulated in experimental settings to convince people to vote by telling them that “everybody does so and so should you” or that voting is desirable, and abstaining is not (Gerber & Rogers, 2009; Murray & Matland, 2014, among others). The conclusion is that both norms matter as the absence of a descriptive norm may impair the effectiveness of the injunctive appeal (Rogers, Fox, & Gerber, 2013). Yet, besides highlighting the importance that descriptive norms should be consistent with injunctive norms, no study has addressed the relative importance of the two types of norms² nor has distinguished pressure coming from friends and pressure coming from the household.

We aim at exploring the mechanism behind social pressure and voting by distinguishing between partner, relatives, and friends and between types of social norms. Our goal is to understand how much pressure to vote citizens are exposed to concretely, what type of pressure, from whom, and with how much impact. As indicated above, we predict pressure from partners to be more frequent and influential than that from friends. We do not have clear expectations about what kind of pressure, descriptive or injunctive, matters most.

Research design

The data

We aim to disentangle the effect of different social agencies using an original data set from Canada including questions on social pressure and voting behavior. This data set is a two-wave Internet survey conducted by YouGov Polimetrix in the provinces of British Columbia and Quebec. The sampling frame was designed to match the demographic makeup of the two provinces as well as expected general political interest (as indicated by the Canadian Election Study). The fieldwork for the first wave was conducted during the last week of the October 2008 Canadian federal election. The sample size was around 2,000 respondents in each province. The second wave took place at the time of the subsequent provincial election, which occurred in December 2008 in the province of Quebec and in May 2009 in British Columbia. The first wave respondents were contacted again for the second wave; about half of them responded to the second wave in each province. The questionnaires were almost identical in the two waves, which allow us to check the robustness of our findings in two separate elections and in two separate provinces.³ The analysis focuses on the first wave, dealing with the federal election (and based on a larger sample) but we show that the findings are similar with respect to the provincial elections (Wave 2).

Measuring social pressure

Previous survey research has employed three different approaches to detect social pressure with regard to voting. The first just asks the individual whether she have been “pressured” in some way by her social environment, which is consistent with injunctive social norms. Questions on whether family or friends have asked, told or encouraged individuals to vote, or fulfill their civic duty have the advantage of being explicit and straightforward.⁴

There is a more indirect approach to injunctive pressure, whereby respondents are asked whether they think that their friends/family would like them to vote and whether they would disapprove their abstention. This is the methodology employed by Abrams, Iversen, and Soskice (2011), who use the following question: “If I didn’t vote, my friends or family (or co-workers) would: 1. Disapprove, 2. Not care, or the topic would never come up.”⁵ The assumption is that those who believe that their friends would disapprove their decision to abstain have detected some cues from their networks indicating what is the desirable behavior (voting), and they foresee some kind of sanction if they don’t vote, which is consistent with injunctive norms. Finally, it is odd that the response categories refer to different concepts (disapprove and care).

The third approach is even more indirect. The idea is that those who notice that their friends and family systematically vote are prone to do the same. This is the approach followed by Cutts and Fieldhouse (2009) and Fieldhouse and Cutts (2012, 2016), who show that living in a household with other people who vote enhances the propensity to vote. This last approach is consistent with descriptive social norms, and it assumes that an individual who knows that her friends are going to vote infers that voting is desirable and behaves in that way.

As we indicate below, the survey we utilize has been inspired by these three approaches.

The models

We wish to determine whether (and how much) different types of social pressure affect the propensity to vote. The dependent variable is the intention to vote. The central independent variables, types of pressure, are explained below.

We wish to ascertain the independent impact of social pressure, and we, therefore, need to control for all the antecedent variables, both sociodemographic characteristics and attitudinal variables, that may influence both the propensity to vote and the likelihood of being exposed to pressure. We include a host of control variables to better isolate the specific effect of social pressure. We also look separately at those with and without partners because these two groups are not exposed to the same set of pressures, that is, the latter group is exempt from pressure from a partner.

Controls include sociodemographic variables. Age and education are the two strongest correlates of turnout (Blais, 2000; Nevitte, Blais, Gidengil, & Nadeau, 2009). We also include gender, religiosity, income, home ownership, whether the person was born in Canada as well as a dummy variable for the province of Quebec (British Columbia is the reference category).

We incorporate all the available attitudes that have been argued to be the most powerful motivational factors that shape people’s propensity to participate in an election (Blais, 2000; Blais & Labbé-St-Vincent, 2011; Brady et al., 1995; Milbrath & Goel, 1977; Rosenstone & Hansen, 1993; Smets & Van Ham, 2013), mainly the belief that voting is a civic duty, one’s general level of interest in politics, and partisanship. Other relevant attitudes are caring about the outcome of the election, and the perceived cost of voting, which are strong predictors of turnout (Blais & Daoust, 2017). All variables except age are recoded from 0 to 1 so we can compare their effects across models. We perform analyses comprising all respondents, and then successively including only people with and without a partner.

The findings

The dependent variable is the intention to vote, as reported in the last week of the campaign. We distinguish those who are certain to vote from those who are not certain.⁶ All in all, as Table 1 presents, 72% of the respondents indicated that they were certain to vote in the federal election.⁷

Table 1.

Descriptive statistics YouGov Canada 2008–2009.

Variable	N	Mean	Standard deviation
Intention to vote	3745	0.716	0.451
Voting friend	3745	0.785	0.411
Voting partner	3745	0.418	0.493
Caring friend	3745	0.109	0.311
Caring partner	3745	0.216	0.412
Quebec province	3745	0.481	0.500
Partner	3745	0.611	0.488
Friends	3745	0.318	0.466
Female	3745	0.484	0.500
Age	3745	45.800	15.722
Education	3745	0.221	0.415
Income	3745	0.427	0.306
Religiosity	3745	0.267	0.299
Home ownership	3745	0.408	0.492
Born in Canada	3745	0.108	0.310
Interest in politics	3745	0.637	0.294
Duty	3745	0.617	0.432
Care	3745	0.764	0.291
Cost	3745	0.865	0.230
Partisanship	3745	0.460	0.499
Altruism	3745	0.756	0.184

The questionnaire includes four “pressure” variables, two about friends and relatives and two about the partner/spouse, two referring to a descriptive norm and two to an injunctive norm. The “descriptive” questions simply ask whether their partner will certainly vote (i.e., a preelection survey) and whether they think that most of their friends and relatives will vote (see Appendix 1 for exact question wordings). The “injunctive” questions ask whether they think that the partner, friends, and relatives care whether they vote or not. This corresponds to the second indirect approach that we distinguished above, whereby we want to see whether people believe that their network would like them to vote (whether they care), the implication being that they feel at least some implicit pressure to vote. This is of course a soft form of pressure. We assume that only those who think that their network cares a lot feel some pressure.

We make the distinction between those living with and without a partner or spouse and those having many friends or not⁸; 61% of the sample lives with a partner and 32% have many friends (see Table 1); 68% of those with a partner believe that their partner will certainly vote, while 81% think that most of their friends and relatives will vote; 35% of those with a partner (and 22% of the total sample) report that their partner cares a lot whether they vote, while 11% of all respondents think that their friends or relatives care a lot.⁹

Table 2 shows the results of the multivariate (logistic) estimation, first for the entire sample and then separately for those with and without a partner. Controls exert the expected influence. One must be careful in the interpretation of the findings as the two pressure variables for the partner are subsets of the “partner” variable. As we are controlling for the presence of a partner, the coefficient associated with “voting partner” and “caring partner” refers to the impact of having a partner who is voting rather than abstaining or caring rather than not caring, among those with a partner. Similarly, the coefficient associated with the “partner” variable refers to the impact of having a partner, controlling for whether that partner votes and cares. The partner variable thus represents the effect of having a partner who does not vote or care. Things are different for friends as one may have few friends but still feel pressure from them to vote. Thus, the two pressure variables for friends and relatives are not subsets of the “friend” variable.

Table 2.

Effects of social pressure on the propensity to vote.

	Model 1—All respondents	Model 2—With partner	Model 3—Without partner
Voting friends	0.67*** (0.11)	0.52*** (0.16)	0.81*** (0.16)
Voting partner	1.91*** (0.15)	1.92*** (0.15)
Caring friends	0.46* (0.20)	0.19 (0.27)	0.75* (0.31)
Caring partner	0.09 (0.17)	0.14 (0.18)
Quebec province	0.61*** (0.11)	0.56*** (0.14)	0.67*** (0.16)
Partner	−1.09*** (0.12)
Friends	−0.24* (0.11)	−0.26 (0.15)	−0.23 (0.16)
Female	0.21* (0.10)	−0.02 (0.14)	0.48** (0.16)
Age	0.01* (0.00)	0.01 (0.00)	0.01 (0.00)
Education	0.31* (0.13)	0.23 (0.17)	0.39 (0.20)
Income	0.19 (0.19)	0.27 (0.25)	0.09 (0.31)
Religiosity	−0.07 (0.17)	−0.15 (0.23)	0.01 (0.25)
Home ownership	−0.28* (0.11)	−0.34* (0.15)	−0.22 (0.15)
Born in Canada	−0.05 (0.16)	0.16 (0.22)	−0.29 (0.24)
Interest in politics	1.56*** (0.20)	1.43*** (0.27)	1.73*** (0.31)
Duty	1.46*** (0.11)	1.42*** (0.15)	1.46*** (0.17)
Care	1.69*** (0.20)	1.58*** (0.28)	1.79*** (0.29)
Cost	2.01*** (0.21)	1.90*** (0.29)	2.13*** (0.29)
Partisanship	0.18 (0.10)	0.16 (0.14)	0.20 (0.16)
Altruism	−0.12 (0.28)	0.06 (0.38)	−0.27 (0.41)
Constant	−4.91*** (0.35)	−5.50*** (0.50)	−5.26*** (0.52)
Pseudo R²	0.376	0.403	0.339
N	3745	2289	1456

Note. Standard errors are in parenthesis.

*p < .05; **p < .01; ***p < .001.

The most important findings concern the four pressure variables: the “voting” partner, “voting” friends, “caring” partner, and “caring” friends.¹⁰ The greatest effect is for whether one is certain that her partner will vote. Everything else being equal, the predicted probability of voting is 23 percentage points higher among those with a voting partner than among those with a nonvoting partner, although an equally interesting comparison is with those without a partner, and then the difference is 11 percentage points.¹¹ Surprisingly, whether the partner cares or not whether the person will vote does not appear to matter. The mere presence of a voting partner suffices to make people vote. It is important to keep in mind that we are controlling for a whole series of powerful individual-level determinants of turnout.¹² It appears that the descriptive norm matters more than the injunctive one.¹³

Things are different with respect to friends. Their impact is due to some extent to the presence of an injunctive norm. As such the presence of voting friends increases the likelihood of voting by 9 percentage points but having friends who care whether the person votes has an additional positive effect of 5 points. It is important to point out, however, that only 11% of the sample indicates that their friends care whether they vote. When there is pressure, it appears to have some effect, but the fact is that pressure from friends is scarce.

We then replicate the analyses separately for those without and with partners (Columns 2 and 3 of Table 2). In the former group, there is no partner and we can thus only examine the impact of friends. Again, we see that both having friends who vote and having friends who care whether we vote have additional independent effects on the propensity to vote. These effects are slightly larger than in the whole sample,¹⁴ as the importance of friends is somewhat greater in the absence of a partner. Things are quite different in the case of those living with a partner. They seem to be impervious to injunctive social pressure, but the presence of a voting partner appears to make a big difference, as it increases the propensity to vote by 26 percentage points. The presence of voting friends has a weaker effect (plus 6 percentage points).

We run a series of additional estimations to ascertain the robustness of our findings and their external validity (see Table 3). We first performed the same estimations with the second wave of the study (conducted at the time of the provincial election, that is, December 2008 in Quebec and May 2009 in British Columbia), predicting the intention to vote in the provincial election (Model 4). The results are very similar, except for the caring friend which is no more significant. The caring partner variable is again not significant, while having a voting partner is clearly the most powerful predictor. Then, we reproduce the Wave 2 estimations while controlling for reported turnout in the previous federal election and pressure variables measured in Wave 1 (Model 5). As expected, previous turnout is significant with a positive effect. The voting partner and voting friends variables remain significant with similar effects. We also checked whether the impact of our four central variables is different in Quebec (QC) and British Columbia, adding four interactions with province dummies (Model 6). None of these interactions are significant. Hence, partner and friends play similarly in these two provinces. Finally, we used the original (six-category) coding of the dependent variable (from certain that will not vote to certain to vote) and an ordered logit estimation (Model 7). The patterns are very much the same as in Table 2 (Column 1). The most powerful factor is still the presence of a voting partner and whether that partner cares about the person voting does not matter. The impact of friends is smaller, but in their case both the presence of voting friends and the fact that they care seem to make a difference.

Table 3.

Effects of social pressure on the propensity to vote: Robustness checks.

	Model 4	Model 5	Model 6	Model 7
	Wave 2	W2 + W1 controls	QC interactions	Ordered logit
Voting friends W2	0.70*** (0.18)	0.61** (0.20)
Voting partner W2	2.25*** (0.22)	2.30*** (0.25)
Caring friends W2	0.32 (0.27)	0.39 (0.31)
Caring partner W2	−0.15 (0.25)	−0.13 (0.29)
Previous vote		2.03*** (0.18)
Voting friends W1		0.12 (0.21)	0.61*** (0.15)	0.30** (0.09)
Voting partner W1		−0.36 (0.24)	1.69*** (0.19)	1.24*** (0.12)
Caring friends W1		−0.32 (0.28)	0.55* (0.24)	0.26 (0.13)
Caring partner W1		0.05 (0.26)	0.24 (0.22)	−0.00 (0.12)
Voting friends # QC			0.11 (0.22)
Voting partner # QC			0.46 (0.26)
Caring friends # QC			−0.34 (0.43)
Caring partner # QC			−0.31 (0.36)
QC province	0.22 (0.15)	0.21 (0.17)	0.46* (0.20)	0.31*** (0.08)
Partner	−1.21*** (0.17)	−1.08*** (0.21)	−1.08*** (0.12)	−0.67*** (0.10)
Friends	−0.22 (0.16)	−0.21 (0.17)	−0.23* (0.11)	−0.04 (0.08)
Control variables not shown
Constant	−5.67*** (0.50)	−6.20*** (0.55)	−4.83*** (0.36)	^a7.51*** (0.30)
Pseudo R²	0.386	0.452	0.377	0.101
N	1773	1743	3745	3745

Note. Standard errors are in parenthesis.

^aConstant for cut 5, ordered logit.

*p < .05; **p < .01; ***p < .001.

In sum, our results show that most people are not subjected to strong injunctive pressure. Only 11% of the respondents think that their friends care a lot whether they vote, and only 35% of those living with a partner say that their partner cares a lot. Second, for those living with a partner (61% of the sample), it makes a big difference whether the partner votes or not. Interestingly, it does not matter whether the partner cares. Whether friends vote and care has only a small impact. Third, for those without a partner (39% of the sample), friends make some difference, both whether they vote and whether they care. This effect is smaller than that of the partner, but it is not negligible. All in all, social pressure is most prominent in the household, it is mostly descriptive, and it has a powerful effect when it occurs (which is rare outside the household).

Conclusion

Our review of the literature on social pressure and voting revealed a corpus of experimental works, mostly confined to the U.S., confirming the effect of the threat to sanction—mostly from neighbors—on the likelihood to vote. Research relying on surveys, in turn, has considered several mechanisms and agencies administering social pressure in an electoral setting. We detected a diversity of works concerned with the effects of family or friends and a great diversity regarding the operationalization of these effects. In this study, we fill some gaps in the literature, paying attention to the different ways (injunctive and descriptive norms) by which social agencies affect the turnout decision.

A limitation in previous research was that the results were often based on only one country—the U.S. This study offers new insights using a survey conducted in two different Canadian provinces in three elections. The survey included four different questions to measure the extent of social pressure they were exposed to. We were particularly keen to distinguish pressure associated with descriptive and injunctive norms and pressure coming from within and outside the household. We wanted to estimate the impact of informal social pressure on turnout but also to gage the frequency and intensity of pressure.

We are aware of some limits concerning the sources of the pressure in our study. First, the combination of friends and relatives in the same survey question limits our interpretation of the results as it is impossible to disentangle one group from the other. Nevertheless, our strongest results are clear as they concern the partner’s pressure, which is not mixed with another social agency. Second, we do not tap other potential sources of pressure, such as colleagues or neighbors. Following our findings, we would not expect high levels of pressure to be exerted by those actors, but we believe it is important to take into account the different potential sources of social pressure to vote.

The clearest finding is that most of the successful social pressure comes from within the household, and more specifically is exerted by the partner/spouse. But perhaps the most surprising result (at least for us) is that descriptive norms seem to be even more powerful than injunctive norms. In fact, what seems to matter is whether the partner votes, not whether she cares.

Our study provides a perspective on how to interpret field experiments that have shown the impact of social pressure on turnout. The first (and most obvious) observation is that these experiments are extremely useful to demonstrate the causal impact of the specific type of pressure that is exercised in those studies but that they cannot tell us how frequent or infrequent that kind of pressure is in “real life.” Our data suggest that pressure from outside the household is rare. From that perspective, our findings appear to be at odds with the Gerber, Green, and Larimer’s (2008) experiment. One way of reconciling these contradictory findings would be that strong social pressure from neighbors is influential but that it seldom occurs.

Our findings are fully consistent with those of Nickerson (2008), who reports that the mobilizing effect of GOTV experiments is substantially transferred to the partner. This is correctly interpreted by Nickerson as a contagion effect. It is not clear, however, whether this contagion effect reflects direct pressure or not. The data used in this study suggest that this may be more “descriptive” contagion than direct pressure. As Fieldhouse and Cutts (2012) aptly put it, this may be just a “companion” effect (see also Glaser, 1959). Whatever the case, the implication of our study is clear. If we wish to make sense of how social pressure affects the decision to vote or abstain, we should look first and foremost at the pressure that does or does not take place in the household.

In this article, like in practically all studies, we have focused on the positive effects of social pressure. But there may well be some subcultures where individuals are encouraged not to vote, and that this could particularly the case in countries with compulsory voting. Whether this is the case is an intriguing question that would require further research.

We end with a plea for the necessity to include a rich battery of questions to adequately tap the frequency and intensity of pressure that people are subject to when it comes to deciding whether to vote or not. It is crucial to distinguish descriptive and injunctive pressure as well as to specify the source of that pressure. It is, therefore, insufficient to include only one or two questions, especially if we want to provide an accurate description of how much or little pressure people are exposed to. And how can we make sense of how much impact social pressure has if we know precious little about how much pressure people are concretely exposed to?

Footnotes

Authors’ Note

Carol Galais is now affiliated with Universitat Autònoma de Barcelona, Spain.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Maxime Coulombe

Open research statement

As part of IARR’s encouragement of open research practices, the authors have provided the following information: This research was not pre-registered. The data used in the research are available. The data can be obtained by emailing andre.blais@umontreal.ca.

Notes

Appendix 1

Table 1A.

Question wording and variable coding.

Variable	Question formulation
Intention to vote	How likely are you to vote in this federal election? Coded 1 if: 1) Certain to vote/I have already voted Coded 0 if: 2) Very likely to vote. Somewhat likely to vote. 4) Somewhat unlikely to vote. 5) Very unlikely to vote. 6) Certain that will not vote.
Previous vote	Did you vote in the last federal election, in October 2008? Coded 1 if: 1) Yes Coded 0 if: 2) No. 3) I don’t remember.
Gender (female)	Are you male or female? Coded 1 if female. Coded 0 if male.
Age	In what year were you born? Coded: 2008—answer (year)
Education	What is the highest level of education that you have completed? Coded 1 for completed BA, MA, PhD or an equivalent. Coded 0 for other answers
Income	What is your household total annual income? Coded from 0 to 1 with six different possible answers, going from 1) less than $20,000 to 6) Over $100,000.
Religiosity	How often do you attend formal religious services? Coded from 0 to 1 with four different possible answers, going from 1) Once a week or more to 4) Never. Respondents who reported having no religion at the question “What is your religion?” are also coded 0.
Home ownership	Do you, or someone else in your household own your home? Coded 1 if: 1) Yes. Coded 0 if: 2) No/rent.
QC province	A dummy variable that equals 1 for Quebec respondents and 0 for BC respondents
Friends	Would you say that you have many or few friends? Coded 1 if: 1) Many. Coded 0 if: 2) Some. 3) Few. 4) Very few.
Born in Canada	Were you born in Canada? Coded 1 if: 1) Yes. Coded 0 if: 2) No.
Interest in politics	In general, how interested are you in federal politics? Use a 0 to 10 scale where 0 means not interested at all and 10 means extremely interested. The scale is recoded from 0 to 10 to 0 to 1.
Duty	The duty variable is constructed from a combination of two questions. Question 1: Different people feel differently about voting. For some voting is a DUTY. They feel that they should vote in every election however they feel about the candidates and parties. For some, voting is a CHOICE. They feel free to vote or not to vote in an election depending on how they feel about the candidates and parties. For you personally, is voting first and foremost a duty or a choice in a federal election? Question 2: How strongly do you feel personally that voting is a duty? Coded 0 if voting is a choice at the first question. Otherwise, it is coded from 0.33 to 1 with three different possible answers (second question), going from 3) Not very strongly to 1) Very strongly.
Partisanship	Generally speaking, do you feel close to one of the federal parties? Coded 1 if: 1) Yes. Coded 0 if: 2) No.
Care	How much do you personally care which party will form the government after the election? Coded from 0 to 1 with four different possible answers, going from 1) Not at all to 4) A lot.
Cost	For you personally, on a scale from 0 to 10, where 0 means very difficult and 10 means very easy, how easy or difficult is it to do the following? -To go the polling station. The scale is recoded from 0 to 10 to 0 to 1.
Altruism	The altruism variable is constructed from a combination of two questions. Question 1: Do you strongly agree, somewhat agree, somewhat disagree, or strongly disagree with the following statements: One should always find ways to help others less fortunate than oneself. Question 2: There are different views about what it means to be a good citizen. For you personally, how important is it for a good citizen to do the following things. Use a scale from 0 to 10 where 0 means not at all important and 10 extremely important. Help people who are worse off than oneself. Score for the first question goes from 0 (strongly disagree) to 1 (strongly agree). Score for the second question is recoded from the 0 to 10 scale to 0 to 1. The variable is coded by using the mean score of the two variables.
Partner	Do you have a partner or spouse? Coded 1 if: 1) Yes. Coded 0 if: 2) No.
Caring partner	Does he/she care whether you vote or not? Coded 1 if: 1) He/she cares a lot. Coded 0 if: 2) He/she cares somewhat. 3) He/she cares a little. 4) He/she does not care at all. 9) I have no idea.
Voting partner	Do you know whether your partner/spouse will vote in this election? Coded 1 if: 1) He/she will certainly vote. Coded 0 if: 2) He/she will probably vote. 3) He/she will probably not vote. 4) He/she will certainly not vote. 9) I have no idea.
Caring friends	Do you think that most of your friends and relatives care whether you vote or not? Coded 1 if: 1) They care a lot. Coded 0 if: 2) They care somewhat. 3) They care a little. 4) They do not care at all. 9) I have no idea.
Voting friends	Do you think that most of your friends and relatives will or will not vote in this election? Coded 1 if: 1) Most of them will vote Coded 0 if: 2) Most of them will not vote. 9) I have no idea.

References

Abrams

Iversen

Soskice

(2011). Informal social networks and rational voting. British Journal of Political Science, 41, 229–257.

Achen

Blais

(2016). Intention to vote, reported vote, and validated vote. In Elkink

J. A.

Farrell

D. M.

(Eds.), The act of voting: Identities, institutions, and locale (pp. 195–209). London: Routledge.

Aquilino

W. S.

(1997). From adolescent to young adult: a prospective study of parent-child relations during the transition to adulthood. Journal of Marriage and the Family, 59, 670–686.

Berelson

B. R.

Lazarsfeld

P. F.

MacPhee

W. N.

(1954). Voting. Chicago, IL: University of Chicago Press.

Berger

(2009). Political theory, political science and the end of civic engagement. Perspectives on Politics, 7, 335–350.

Bhatti

Hansen

K. M.

(2012). Leaving the nest and the social act of voting: turnout among first-time voters. Journal of Elections, Public Opinion and Parties, 22, 380–406.

Blais

(2000). To vote or not to vote?: The merits and limits of rational choice. Pittsburgh: University of Pittsburgh Press.

Blais

Achen

. (forthcoming). Civic duty and voter turnout. Political Behavior. doi:10.1007/s11109-018-9459-3

Blais

Daoust

J.-F

. (2017). Do I wish to vote? Montreal: Mimeo.

10.

Blais

Labbé-St-Vincent

(2011). Personality traits, political attitudes and the propensity to vote. European Journal of Political Research, 50, 395–417.

11.

Bond

R. M.

Fariss

C. J.

Jones

J. J.

Kramer

A. D.

Marlow

Settle

J. E.

Fowler

J. H.

(2012). A 61-million-person experiment in social influence and political mobilization. Nature, 489, 295–298.

12.

Brady

H. E.

Verba

Schlozman

K. L.

(1995). Beyond SES: a resource model of political participation. American Political Science Review, 89, 271–294.

13.

Brown

B. B.

Clasen

D. R.

(1985). Peer pressure inventory. Journal of Youth and Adolescence, 14, 451–468.

14.

Buhl

H. M.

(2009). My mother: My best friend? Adults’ relationships with significant others across the lifespan. Journal of Adult Development, 16, 239.

15.

Campbell

D. T.

(1982). Legal and primary-group social controls. Journal of Social and Biological Structures, 5, 431–438.

16.

Carbery

Buhrmester

(1998). Friendship and need fulfillment during three phases of young adulthood. Journal of Social and Personal Relationships, 15, 393–409.

17.

Cialdini

R. B.

Reno

R. R.

Kallgren

C. A.

(1990). A focus theory of normative conduct: recycling the concept of norms to reduce littering in public places. Journal of Personality and Social Psychology, 58, 1015.

18.

Collins

van Dulmen

(2006). Friendships and romance in emerging adulthood: assessing distinctiveness in close relationships. In Arnett

J. J.

Tanner

J. L.

(Eds.), Emerging adults in America: Coming of age in the 21st century (pp. 219–234). Washington, DC: American Psychological Association.

19.

Cutts

Fieldhouse

(2009). What small spatial scales are relevant as electoral contexts for individual voters? The importance of the household on turnout at the 2001 general election. American Journal of Political Science, 53, 726–739.

20.

Denver

(2008). Another reason to support marriage? Turnout and the decline of marriage in Britain. The British Journal of Politics and International Relations, 10, 666–680.

21.

Downs

(1957). An economic theory of democracy. New York: Harper Collins Publishers.

22.

Etcheverry

P. E.

Agnew

C. R.

(2009). Similarity in cigarette smoking attracts: a prospective study of romantic partner selection by own smoking and smoker prototypes. Psychology of Addictive Behaviors, 23, 632.

23.

Fieldhouse

Cutts

(2012). The companion effect: household and local context and the turnout of young people. The Journal of Politics, 74, 856–869.

24.

Fieldhouse

Cutts

(2016). Shared partisanship, household norms and turnout: Testing a relational theory of electoral participation. British Journal of Political Science, 48, 1–17.

25.

Fowler

J. H

. (2005). 14 Turnout in a Small World. In Zuckerman

A. S.

(Ed.), The social logic of politics: Personal networks as contexts for political behavior (pp. 269–287). Philadelphia: Temple University Press.

26.

Garcia Bedolla

Michelson

M. R

. (2012). Mobilizing inclusion: Transforming the electorate through get-out-the-vote campaigns. New Haven, CT: Yale University Press.

27.

Gerber

A. S.

Rogers

(2009). Descriptive social norms and motivation to vote: everybody’s voting and so should you. The Journal of Politics, 71, 178–191.

28.

Gerber

A. S.

Green

D. P.

Larimer

C. W.

(2008). Social pressure and voter turnout: evidence from a large-scale field experiment. American Political Science Review, 102, 33–48.

29.

Gerber

A. S.

Green

D. P.

Larimer

C. W.

(2010). An experiment testing the relative effectiveness of encouraging voter participation by inducing feelings of pride or shame. Political Behavior, 32, 409–422.

30.

Glaser

W. A.

(1959). The family and voting turnout. Public Opinion Quarterly, 23, 563–570.

31.

Goerres

(2007). Why are older people more likely to vote? The impact of ageing on electoral turnout in Europe. The British Journal of Politics & International Relations, 9, 90–121.

32.

Green

D. P.

Gerber

A. S

. (2015). Get out the vote: How to increase voter turnout. Washington, DC: Brookings Institution Press.

33.

Guan

S. S. A.

Fuligni

A. J.

(2016). Changes in parent, sibling, and peer support during the transition to young adulthood. Journal of Research on Adolescence, 26, 286–299.

34.

Huckfeldt

Sprague

(1995). Citizens, parties, and social communication. New York: Cambridge.

35.

Kenny

(1993). The microenvironment of political participation. American Politics Quarterly, 21, 223–238.

36.

Knack

(1992). Civic norms, social sanctions, and voter turnout. Rationality and Society, 4, 133–156.

37.

Knack

(1995). Does “motor voter” work? Evidence from state-level data. The Journal of Politics, 57, 796–811.

38.

Mann

C. B.

(2010). Is there backlash to social pressure? A large-scale field experiment on voter mobilization. Political Behavior, 32, 387–407.

39.

Milbrath

L. W.

Goel

(1977). Political participation: How and why do people get involved in politics? (2nd ed.). Chicago, IL: Rand McNally College Pub. Co.

40.

Murray

G. R.

Matland

R. E.

(2014). Mobilization effects using mail: social pressure, descriptive norms, and timing. Political Research Quarterly, 67, 304–319.

41.

Nevitte

Blais

Gidengil

Nadeau

(2009). Socio-economic status and nonvoting: a cross-national comparative analysis. In Klingemann

H. D.

(Eds.), The comparative study of electoral systems (pp. 85–108). Oxford, UK: Oxford University Press.

42.

Nickerson

D. W.

(2008). Is voting contagious? Evidence from two field experiments. American Political Science Review, 102, 49–57.

43.

Panagopoulos

(2014). Watchful eyes: implicit observability cues and voting. Evolution and Human Behavior, 35, 279–284.

44.

Panagopoulos

Larimer

C. W.

Condon

(2014). Social pressure, descriptive norms, and voter mobilization. Political Behavior, 36, 451–469.

45.

Pattie

Johnston

. (2000). ‘People who talk together vote together’: an exploration of contextual effects in Great Britain. Annals of the Association of American Geographers, 90, 41–66.

46.

Riker

W. H.

Ordeshook

P. C.

(1968). A theory of the calculus of voting. American Political Science Review, 62, 25–42.

47.

Rogers

Fox

C. R.

Gerber

A. S.

(2013). Rethinking Why People Vote: Voting as Dynamic Social Expression. In Shafir

(Eds.), The Behavioral Foundations of Public Policy (pp. 91–107). Pricenton, NJ: Princeton University Press.

48.

Rogers

Green

D. P.

Ternovski

Young

C. F.

(2017). Social pressure and voting: a field experiment conducted in a high-salience election. Electoral Studies, 46, 87–100.

49.

Rosenstone

Hansen

(1993). Mobilization, participation, and American democracy. New York: McMillan.

50.

Scheepers

Van der Slik

(1998). Religion and attitudes on moral issues: effects of individual, spouse and parental characteristics. Journal for the Scientific Study of Religion, 37, 678–691.

51.

Sheeran

Orbell

(1999). Augmenting the theory of planned behavior: roles for anticipated regret and descriptive norms. Journal of Applied Social Psychology, 29, 2107–2142.

52.

Sinclair

(2012). The social citizen: Peer networks and political behavior. Chicago, IL: University of Press.

53.

Smets

Van Ham

(2013). The embarrassment of riches? A meta-analysis of individual-level research on voter turnout. Electoral Studies, 32, 344–359.

54.

Stice

Ziemba

Margolis

Flick

(1996). The dual pathway model differentiates bulimics, subclinical bulimics, and controls: testing the continuity hypothesis. Behavior Therapy, 27, 531–549.

55.

Stoker

Jennings

(1995). Life-cycle transitions and political participation: the case of marriage. American Political Science Review, 89, 421–433.

56.

Straits

B. C.

(1990). The social context of voter turnout. Public Opinion Quarterly, 54, 64–73.

57.

Wolfinger

N. H.

Wolfinger

R. E.

(2008). Family structure and voter turnout. Social Forces, 86, 1513–1528.

58.

Wolfinger

R. E.

Rosenstone

S. J.

(1980). Who votes? (Vol. 22). New Haven, CT: Yale University Press.