A sociocultural approach to voting: Construing voting as a duty to others predicts political interest and engagement

Significance

Between one-third and one-half of eligible voters in the United States fail to turn out for elections. We take a sociocultural approach to understand this lack of turnout, examining how people construe—or make sense of—the duty to vote. Our results suggest that construing voting as a duty to others (i.e., as interdependent)—whether spontaneously or experimentally—is associated with feeling an increased obligation to vote. Perceiving voting as an obligation, in turn, predicts people’s intentions to be politically engaged in the future. The present work points to the potential of construal-based interventions as a scalable complement to existing strategies aimed at increasing political engagement.

Abstract

A representative democracy requires citizens to be politically engaged; however, a substantial portion of eligible United States voters do not vote. While structural (e.g., ease or difficulty of voting) and individual (e.g., political efficacy, civic knowledge) factors contribute to (a lack of) turnout, the present work adopts a sociocultural perspective to investigate an additional contributor: how people construe—or make sense of—the duty to vote. We examine whether, and for whom, construing voting as interdependent (i.e., voting as a duty to others), compared to independent (i.e., voting as a duty to self), is associated with increased perceived duty and political engagement. Archival analysis (n$=$ 10,185) documents how perceived duty to vote relates to voter turnout in a nationally representative sample of Americans (Study 1). Two preregistered studies (total n$=$ 1,256) provide evidence that naturalistically construing one’s duty to vote as interdependent (Study 2) and experimentally reflecting on interdependence (Study 3) both predict increases in perceived voting duty. Perceived duty to vote, in turn, is associated with heightened political engagement intentions. Taken together, the present work suggests that how voting is construed—as an independent duty to the self or an interdependent duty to others—may meaningfully influence political engagement, with implications for voter turnout interventions.

Voting is central to democracy. In theory, the ballot box is an equalizer; citizens who vote, regardless of their demographic background, are better able to ensure government representation (1–3). Yet nearly one-third of eligible voters in the United States do not vote in general elections, a proportion that increases to one-half for midterms (4).

Turnout differs as a function of social categories, including social class, race, age, and their intersections (5–11). Notably, there is a persistent social class gap in turnout: People from working-class contexts (i.e., those who have not attained a 4-y college degree) are less likely to vote relative to those from middle-class contexts (5–9). US Census Bureau data from the 2000 presidential election reveal a 22.6 pt difference in turnout, with 49.4% of individuals from working-class contexts voting compared to 72% of their middle-class counterparts. Two decades later, the class-based gap in turnout was remarkably similar: 51.1% of individuals from working-class contexts voted in the 2020 presidential election compared to 74.2% of those from middle-class backgrounds (a 23.1 pt difference). This lack of political engagement is consequential because the preferences and policies of large constituencies of eligible voters remain considerably underrepresented in the American government (12, 13). Given this lack of turnout, both in general and especially for those from working-class contexts, there is room to boost political interest and engagement among potential voters.

Structural (e.g., access to polling places) and individual (e.g., motivational) factors both contribute to (a lack of) turnout. Relative to their middle-class counterparts, those from working-class contexts are especially likely to contend with higher structural barriers to voting. They are more likely to depend on public transportation (14), are less likely to receive time off from work (15), and may experience longer wait times at polling places (16), all of which increase the effort required to vote. When voting is costly, people are less politically engaged (17, 18) even when their political motivation is high (19). Barriers to voting can also arise from individual-level factors, such as political efficacy and civic duty (20). Relative to individuals from middle-class contexts, those from working-class contexts report feeling less political efficacy (21) and lower perceived voting duty (22, 23), both of which are associated with reduced political interest and engagement (24, 25).

While the aforementioned factors undoubtedly influence voting, here we theorize and document that sociocultural factors reflect an additional contributor. We propose that how voting is construed—as an independent duty to the self or an interdependent duty to others—may meaningfully influence political engagement. We theorize that interdependent (vs. independent) construals of voting duty will be associated with increased perceived voting obligation. Specifically, we test whether, and for whom, construing the duty to vote as interdependent is associated with heightened perceived voting duty to ultimately aid in the development of effective and scalable voter turnout interventions.

A Sociocultural Perspective on the Duty to Vote

People who perceive having a duty or an obligation to vote are more likely to be politically engaged than those who do not (24, 25). Perceived duty tends to differ along class lines: Those from middle-class contexts are more likely, on average, to perceive having a duty to vote compared to their working-class counterparts (22, 23). However, even among individuals from middle-class contexts, there is not a full consensus that voting is a duty. Further, this existing work demonstrating class-based gaps may imply stable, enduring individual differences in whether people perceive voting as a duty—differences arising from childhood socialization processes (23). An implication is that shifting people’s perceived voting duty may require changes in childhood.

Here, we examine perceived voting duty through a sociocultural lens. A sociocultural approach suggests that how people understand themselves and their behavior is shaped by the context in which that behavior occurs: The same situation or action can be understood—or construed—in multiple ways (26, 27). We propose that people’s perceptions of voting duty and subsequent political engagement may stem, in part, from how people construe the duty to vote (i.e., as an independent obligation to self or an interdependent obligation to others). We therefore investigate whether a sociocultural approach adds to our understanding of voter turnout, both in general and as a function of social class.

Voting as an Independent Duty.

For many U.S. institutions (e.g., higher education, corporate workplaces), independent construals abound; people’s behaviors are viewed as self-focused actions arising from individual interests and goals (28–30). Anecdotal evidence suggests that political behavior might likewise tend to be understood in terms of independence.^* The front page of the website for When We All Vote, an organization created by Michelle Obama to encourage turnout, emphasizes the importance of voting to “ensure your voice is heard” (https://whenweallvote.org/). The League of Women Voters, a turnout organization that has existed for over a century, urges people to “support legislation that stands up for your values” (https://www.lwv.org/). Rock the Vote, a popular turnout and political education organization, states, “A healthy democracy requires active participation and prepares its people to self-govern by providing a practical understanding of how government works, how power is leveraged, and the skills to exercise their rights and represent their interests” (https://www.rockthevote.org/). The language around making one’s voice heard, standing up for one’s values, and representing one’s interests is consistent with construing the duty to vote as independent (i.e., a duty to self whereby people have an individual duty to make their voice heard, express their opinions, and exercise their right).

Voting as an Interdependent Duty.

We suggest that the duty to vote can also be understood or construed in terms of interdependence (i.e., a duty to others whereby people have a duty to support others, respond to the needs and interests of others, and contribute to collective change). While people may visit the ballot box independently, voting is inherently a cooperative and other-focused behavior (e.g., for any given individual, the costs of voting outweigh immediate benefits; only when many people vote can voting benefit the collective). Research suggests that when people perceive or construe behaviors as interdependent, they are more likely to cooperate with others (31). Thus, we theorize that the extent to which people construe the duty to vote as interdependent—whether naturalistically or experimentally—will be positively associated with perceived obligation to vote.

Though not applied to voting, prior work has examined the ways interdependent construals shape people’s psychology and behavior in other consequential domains like higher education. Findings from cultural and social psychology provide evidence that interdependent construals might uniquely benefit people from working-class contexts by buffering against social identity threat and creating a cultural match with interdependent working-class norms and values (30, 32). In contrast, other work suggests that interdependent construals confer universal benefits across social class contexts, boosting perceived obligation to others and intrinsic motivation for all (33–36).

Integrating and extending this research to the political domain, we examine perceived voting duty in a nationally representative sample of Americans. Replicating prior work, we find that people from working-class (vs. middle-class) contexts are less likely to perceive having a duty to vote (23). In addition, we find that disparities in perceived voting duty help explain class-based gaps in turnout above and beyond individual and structural controls. While our results suggest that those from middle-class (vs. working-class) contexts are more likely to perceive voting as a duty, we nevertheless find that a substantial portion of individuals from middle-class contexts (nearly 40%) do not perceive voting as such.

Building on these findings, we outline two accounts regarding who will benefit from interdependent (vs. independent) construals of voting duty. On the one hand, construing voting as an interdependent duty may uniquely benefit individuals from working-class contexts as it aligns more closely with working-class norms and values (30, 37). This possibility would help explain social class gaps in perceived voting duty observed in our nationally representative data; if behaviors like voting tend to be understood in independent terms, this is especially misaligned with interdependent, working-class norms. On the other hand, given the fundamentally interdependent nature of voting and failures to turn out across the social class spectrum, an emphasis on interdependence may boost perceived voting duty and political interest for all. We outline these two accounts in greater detail below.

Creating a Cultural Match: The “Unique Working-Class Benefits” Account

While individuals from middle-class contexts in the United States tend to understand themselves and their situations in terms of independence, people from working-class contexts are more likely to be guided by interdependent norms (30, 37). A person behaving in line with interdependent norms will follow rules, adjust to situations, and respond to the needs and interests of others (28, 29, 38). Applied to the political context, construing voting as an interdependent duty may uniquely benefit those from working-class contexts as it aligns with their more interdependent norms and values.

A match between people’s cultural values and the values reflected in the broader context can reduce psychological discomfort and enhance feelings of belonging, prompting increased engagement (30, 39, 40). Values affirmation, a widely studied strategy intended to reduce discomfort and increase this perceived alignment, operates by prompting people to reflect on their values to buffer against identity threats produced by challenging and effortful performance situations. For example, while higher education in the United States tends to be characterized by independent norms, when given the opportunity to reflect on interdependent values, students from working-class contexts show increased academic engagement (41), while those from middle-class contexts are unaffected.^† If behaviors in the political domain likewise tend to reflect independence, this may be especially misaligned with interdependent, working-class norms and values.

Directly reframing independent-seeming actions as interdependent can similarly prompt increased engagement for those from working-class contexts (30, 32). For example, people tend to construe power acquisition as an independent, self-focused action (i.e., gaining power via self-promotion), leading those from working-class (vs. middle-class) contexts to exhibit less interest in obtaining it. When attaining power is instead reframed as an interdependent, other-focused action (i.e., gaining power by behaving in the group’s interest), people from working-class contexts are more willing to pursue it, while those from middle-class contexts are unaffected by the reframing manipulation (32).

Taken together, a “cultural match” (or unique working-class benefits) account suggests that interdependent construals of voting duty—whether spontaneous or experimentally induced—will predict increased perceived duty and political interest among individuals from working-class (but not middle-class) contexts.

Though interdependent construals are not necessarily aligned with independent, middle-class norms, a cultural match account further predicts that rather than backfiring (42), interdependent construals will not affect individuals from middle-class contexts. As a well-represented group in American politics (12), people from middle-class contexts are less likely to question their belonging and may assume that the cultural norms of their group will be included regardless of how voting is construed (43). In sum, a unique working-class benefits account predicts that the extent to which people from working-class (but not middle-class) contexts construe the duty to vote as interdependent will predict perceived voting duty and political interest (H1).

The “Universal Benefits” Account

An alternative possibility is that interdependent construals of voting duty will confer broad benefits regardless of people’s social class context, predicting increases in perceived voting duty and interest for all. Despite people from middle-class contexts in the United States understanding themselves in terms of independence (37), interdependence is an evolved, fundamental, and universal social motive (44, 45). Interdependence sets humans apart from other species; people are motivated to share goals, intentions, and experiences with others (46), which allows for large-scale cooperative behaviors—behaviors that are individually costly, but increase collective benefits (45). Voting is one such cooperative behavior, costly and effortful at the individual-level and beneficial to the collective.

Evidence outside of the political domain suggests that interdependent construals predict collectivist values and cooperative behaviors for all. For example, perceiving situations as interdependent is associated with cooperation, regardless of people’s social class context (31). And when interdependence (vs. independence) is made salient, people are more likely to construe the self in interdependent terms, subsequently endorsing more collectivist goals (e.g., belongingness, safety) and a greater social obligation to others (36). Moreover, when a task is framed as an interdependent opportunity to work with others, people show increased intrinsic motivation and improved task performance, regardless of their social class context (33–35).

While some work suggests that reflecting on interdependent values increases engagement by reducing belonging threats only for those from working-class contexts (41), other work finds more wide-ranging effects. One study found that an intervention which provided students the opportunity to write about both the traditional independent values of higher education (e.g., Independence, Curiosity, etc.) alongside more interdependent values (e.g., Relationships with Friends and Family, Belonging to a Social Group, etc.) increased academic engagement and performance for all students, regardless of social class context (47). This universal boost came, in part, because it allowed students to reflect on multiple values.

Taken together, a universal benefits account predicts that the extent to which people, regardless of their social class context, construe the duty to vote as interdependent will predict increased perceived duty and political interest (H2).

Overview of Studies

We examine our research questions across three studies (two of which were preregistered; see Materials and Methods or SI Appendix, section 1 for links). Study 1 uses a large, nationally representative archival dataset to examine how perceived voting duty relates to voter turnout, both in general and as a function of people’s social class context. In Studies 2 and 3, we shift to focus on voting as an interdependent duty to provide correlational and experimental tests of H1 and H2. Study 2 leverages a survey to examine whether spontaneous use of interdependent (vs. independent) language when describing one’s duty to vote is associated with increased perceived voting duty and interest only among individuals from working-class contexts (H1) or for all (H2). Finally, Study 3 examines whether experimentally reflecting on interdependence (vs. independence) boosts perceived voting duty and interest only for those from working-class contexts (H1) or for everyone (H2). Together, the present work highlights the importance of construal in explaining people’s perceived duty to vote and points to potential strategies to increase political interest.

Following prior work in cultural and social psychology (48), we use the term working-class (WK) to refer to individuals without a 4-y college degree and use the term middle-class (MD) to refer to those with at least a 4-y college degree. Prior work outlines three key proxies for social class: educational attainment, income, and occupation. Of the three, attainment of a 4-y college degree best predicts health and well-being (49, 50) and is particularly consequential for how people see themselves, others, and the world around them (48, 51).

Study 1

In Study 1 (n$=$ 10,185), we examine how perceived duty to vote relates to validated voter turnout, both in the population as a whole and as a function of people’s social class context. Using data from the American National Election Studies (ANES), we explore perceptions of voting duty, including whether perceived voting duty predicts validated voting while accounting for individual- (e.g., demographic characteristics, attention to politics) and structural-level factors (e.g., state-level measures of voting difficulty). In addition, Study 1 examines whether perceived voting duty helps explain social class gaps in turnout while accounting for meaningful controls.

First, we find that half of survey participants perceive voting as a duty (49.9%), while the remaining half (50.1%) perceive voting as a “choice” or “neither a duty nor a choice.” Perceiving voting as a duty (vs. as a choice or “neither”) is a significant positive predictor of validated voting behavior, b$=$ 0.246, SE $=$ 0.009, 95% [0.228, 0.264], P$<$ 0.001, ${\eta }^2$$=$ 0.071. Examined differently, for those who perceive having a duty to vote, 82.2% turned out to vote. In contrast, only 57.6% of those who perceive voting as a choice or neither turned out to vote (a 24.6 pt difference).^‡ The association between perceived duty and turnout is robust to the inclusion of individual (e.g., social class, attention to politics, political conservatism, gender, race, age, household income) and structural (e.g., state-level difficulty of voting) factors (Table 1).^§ Even when accounting for meaningful covariates, the relationship between perceived voting duty and turnout is significant, b$=$ 0.150, SE $=$ 0.01, 95% [0.130, 0.169], P$<$ 0.001, ${\eta }^2$$=$ 0.059.^¶

Table 1.

Predictors of perceived voting duty, self-reported intention to vote, and validated turnout [95% CI]

	Perceived voting duty	Intention to vote	Validated turnout
Perceived voting duty	—	0.054*** [0.043, 0.064]	0.150*** [0.130, 0.169]
Social class	0.116*** [0.092, 0.140]	0.010 [$-$0.002, 0.021]	0.061*** [0.040, 0.082]
Attention to politics	0.073*** [0.063, 0.084]	0.023*** [0.018, 0.028]	0.046*** [0.037, 0.055]
Political conservatism	0.022*** [0.014, 0.029]	$-$0.001 [$-$0.004, 0.003]	$-$0.003 [$-$0.009, 0.004]
Gender	0.024* [0.002, 0.045]	0.009 [$-$0.001, 0.019]	0.047*** [0.028, 0.066]
Race	0.061*** [0.036, 0.087]	0.001 [$-$0.011, 0.013]	$-$0.035** [$-$0.058, $-$0.012]
Age	0.003*** [0.003, 0.004]	0.001*** [0.0003, 0.001]	0.004*** [0.003, 0.004]
Household income	0.005*** [0.003, 0.006]	0.001*** [0.001, 0.002]	0.007*** [0.006, 0.008]
Difficulty of voting	$-$0.001 [$-$0.017, 0.015]	$-$0.002 [$-$0.009, 0.006]	$-$0.005 [$-$0.019, 0.008]
Intercept	$-$0.154*** [$-$0.221, $-$0.087]	0.778*** [0.745, 0.810]	0.137*** [0.078, 0.195]

Note: *P$<$ 0.05. **P$<$ 0.01. ***P$<$ 0.001. Data come from the 2012 and 2016 ANES Time Series Surveys. Social Class is based on education level, where 0 $=$ no 4-y college degree (working-class) and 1 $=$ 4-y college degree (middle-class). Perceived Voting Duty: 1 $=$ duty, 0 $=$ choice, and neither a duty nor a choice. Gender: 0 $=$ male, 1 $=$ female. Race: 0 $=$ White, 1 $=$ racial minority. Difficulty of voting is measured at the state level and scored such that higher values represent greater voting difficulty. Analyses are weighted using the weight_full variable for the 2012 ANES and the V160102 variable for the 2016 ANES.

Consistent with prior work (23), we find evidence of social class differences in perceived duty to vote; while 61.6% of individuals from middle-class contexts perceive voting as a duty, only 43.5% of those from working-class contexts perceive it as such, ${\chi }^2$(1, 9422) $=$ 282.09, P$<$ 0.001. Moreover, we find that perceived duty helps explain class-based gaps in turnout. Using the “mediation” package in R (52), we test for mediation with social class context (0 $=$ working-class, 1 $=$ middle-class) as the predictor, perceived voting duty as the mediator, and validated voting as the outcome. In addition to a significant direct effect of social class context on voter turnout, b$=$ 0.140, 95% [0.118, 0.160], P$<$ 0.001, and a significant total effect, b$=$ 0.181, 95% [0.158, 0.200], P$<$ 0.001, we observe a significant indirect effect of social class context on validated voting behavior via perceived voting duty, b$=$ 0.041, 95% [0.034, 0.050], P$<$ 0.001, that is robust to the inclusion of controls, b$=$ 0.018, 95% [0.012, 0.020], P$<$ 0.001. These results suggest that, in addition to having direct effects on voting likelihood and on perceived duty to vote, social class indirectly affects turnout by shaping the extent to which people perceive voting as a duty.

In sum, Study 1 suggests that differences in perceived voting duty help explain social class gaps in turnout above and beyond other meaningful factors. While those from working-class (vs. middle-class) contexts are less likely to perceive voting as a duty, rates of perceiving voting as a duty are still far from 100% even among those from middle-class contexts. In Study 2, we examine whether, and for whom, construing one’s duty to vote as interdependent (vs. independent) is associated with heightened perceived obligation to vote.

Study 2

As a correlational test of our alternative accounts, we asked a sample of participants (n$=$ 431) to write why they believe there is a duty to vote and to report their political engagement intentions. Using Linguistic Inquiry and Word Count (53), we coded their responses for interdependent and independent language use as indexed by the validated independence-interdependence dictionary (54). Words in the independent dictionary include themes relating to individual interest and achievement, while words in the interdependent dictionary include themes pertaining to social relationships. Participants then responded to a series of measures intended to gauge their overall perceptions of the duty to vote (a composite measure including perceived duty to vote, interest in voting, and importance of voting). For simplicity, we refer to this composite measure as perceived voting duty in the analyses reported below. Participants then reported their intentions to engage in various political activities.

Analyses using Linguistic Inquiry and Word Count (LIWC) suggest that all participants, regardless of social class context, use significantly more interdependent (vs. independent) language in their responses (M_inter = 7.47%, M_ind = 2.39%, t(430) $=$ 18.88, P $<$ 0.001). It is worth noting, however, that 18.3% of participants use more independent (vs. interdependent) language in their responses, suggesting room to shift construals.^# For the full sample of participants, using more interdependent (vs. independent) language to describe one’s duty to vote is significantly positively associated with perceived voting duty, b$=$ 0.052, SE $=$ 0.013, 95% [0.026, 0.078], P $<$ 0.001, ${\eta }_p^2$$=$ 0.04 (Fig. 1, Top). This association between interdependent (vs. independent) language and our composite measure of perceived voting duty holds when controlling for age, gender, race, social class, political affiliation, income, and past political behavior, b$=$ 0.031, SE $=$ 0.011, 95% [0.009, 0.053], P$=$ 0.007, ${\eta }_p^2$$=$ 0.05.

Fig. 1.

Association between interdependent (vs. independent) language use, perceived voting duty and political intentions (Study 2). Note: Error bars reflect SE. Top panel: full sample. Bottom Left panel: working-class sample. Bottom Right panel: middle-class sample. Political intentions (Avg) consists of 13 actions, including the two shown: voting in general election, voting in primary election. Duty composite (Avg) is a composite measure of duty, importance, and interest.

We next tested for mediation using the R package “lavaan” (55), with interdependent (vs. independent) language as the predictor, perceived duty to vote as the mediator, and political intentions as the outcome. While we observe no direct effect of interdependent language on political engagement intentions, b $=$ 0.002, SE $=$ 0.009, 95% [$-$0.016, 0.020], P$=$ 0.793, this analysis suggests a significant total effect of interdependent (vs. independent) language on political intentions, b$=$ 0.028, SE $=$ 0.011, 95% [0.006, 0.049], P$=$ 0.013, as well as a significant indirect effect of interdependent language on political intentions via the mechanism of perceived voting duty, b$=$ 0.025, SE $=$ 0.007, 95% [0.012, 0.038], P$<$ 0.001.^‖ Thus, supporting the universal benefits account (H2), for all participants, interdependent (vs. independent) language use when describing one’s duty to vote is a significant positive predictor of perceived voting duty and, in turn, political engagement intentions.

Next, to test the unique working-class benefits account (H1), we examine the interaction between social class context and interdependent language use predicting perceived duty to vote. Although we observe no significant interaction (b$=$ 0.041, SE $=$ 0.026, 95% [$-$0.010, 0.092], P$=$ 0.112), because our preregistration indicated we would look separately at those from working- vs. middle-class contexts, we next examine the effects of interdependent (vs. independent) language separately for the subsample of participants from working-class (n$=$ 218) and middle-class (n$=$ 213) contexts.

For individuals from working-class contexts, using more interdependent (vs. independent) language when describing one’s duty to vote is significantly positively associated with perceived voting duty, b$=$ 0.071, SE $=$ 0.019, 95% [0.033, 0.109], P$<$ 0.001, ${\eta }^2$$=$ 0.06 (Fig. 1, Bottom Left). Moreover, this relationship between interdependent (vs. independent) language and perceived voting duty holds when controlling for relevant covariates, b $=$ 0.037, SE $=$ 0.017, 95% [0.003, 0.071], P $=$ 0.034, ${\eta }^2$$=$ 0.08. Using lavaan to test for mediation with interdependent (vs. independent) language use as the predictor, perceived duty to vote as the mediator, and political engagement intentions as the outcome, we observe a significant indirect effect of interdependent language use on political intentions via the mechanism of increased duty perceptions among the sample of individuals from working-class contexts, b$=$ 0.034, SE $=$ 0.010, 95% [0.015, 0.054], P$=$ 0.001. For individuals from middle-class contexts, more interdependent (vs. independent) language use when describing one’s duty to vote is marginally significantly associated with perceived voting duty, b$=$ 0.030, SE $=$ 0.017, 95% [$-$0.003, 0.063], P$=$ 0.072, ${\eta }^2$$=$ 0.02 (this effect is not significant when controls are included, b$=$ 0.018, SE $=$ 0.014, 95% [$-$0.010, 0.047], P$=$ 0.205, ${\eta }^2$$=$ 0.02).

In sum, Study 2 provides initial support for H2, suggesting that among all participants, using more interdependent (vs. independent) language when describing the duty to vote is significantly associated with perceived voting duty, which predicts political engagement intentions. In contrast, results do not support the cultural match account (H1) as interdependence seems to benefit individuals across social class contexts. Nevertheless, while subgroup analyses do not support the notion that interdependence is uniquely beneficial to individuals from working-class contexts, these analyses do tentatively suggest relatively larger relationships between interdependence (vs. independence) and perceived voting duty for individuals from working-class (vs. middle-class) contexts. While interdependent language is more prevalent, we do find that 18% of participants use more independent (vs. interdependent) language when describing the duty to vote, suggesting room to shift construal. To replicate and extend these correlational findings, Study 3 provides an experimental test of the effects of having people reflect on interdependence.

Study 3

In Study 3, we test whether experimentally reflecting on interdependence (vs. independence) boosts perceived voting duty and interest uniquely for individuals from working-class contexts (H1) or for all (H2). Participants (n$=$ 825) were randomly assigned to one of two conditions, in which they were either prompted to reflect on interdependent or independent values. In the interdependent condition, participants selected from a list of interdependent values (e.g., Relationships with Family, Being Loyal to Others, etc.) and wrote about why these values were important to them. In the independent condition, in contrast, participants chose from a list of independent values (e.g., Valuing Personal Accomplishments, Being Unique, etc.).

Rather than directly reframing voting itself, this exercise serves to make interdependent or independent values salient to participants. Building on three streams of prior research, we theorized that reflecting on interdependent (vs. independent) values would boost perceived voting duty. First, given the independent norms characteristic of many U.S. institutions (28–30), perceived default voting norms may likewise reflect independence (see SI Appendix, Study 1 for initial evidence in line with this possibility). However, results from Study 2 suggest that people’s own construals of voting tend to reflect interdependence. Reflecting on interdependent values may therefore alleviate this potential misalignment, bolstering people’s belief that they can effectively contribute to the political domain, thus increasing their engagement (41, 47).

Second, prior work suggests that cues of interdependence boost intrinsic motivation (34) and motivate cooperative behaviors to address large-scale collective action problems, such as proenvironmental behaviors to address climate change (35). Given that political action similarly requires collective cooperation, making interdependence salient may likewise boost political interest.

Finally, previous research suggests that when interdependence is made salient, people are more likely to understand the self in interdependent terms, which increases their perceived obligation to others (36). Extended to the political context, making interdependence salient may increase people’s perceived obligation to vote. While we do not test these mechanisms explicitly, in light of this evidence, we theorized that having participants reflect on interdependent values would boost perceived voting duty. We further test whether this theorizing holds only for those from working-class contexts (H1) or for all (H2).

Following our preregistration (which was designed to examine whether reflecting on interdependence boosts perceived duty for individuals from working-class contexts or middle-class contexts or both), we conduct a two-way ANOVA with experimental condition (interdependent, independent) and participant social class context (working, middle) predicting perceived voting duty (a composite measure identical to Study 2). Controlling for covariates, we find a significant main effect of social class on perceived voting duty, F(1, 731) $=$ 77.812, P$<$ 0.001, ${\eta }_p^2$ $=$ 0.10. Consistent with prior work (23) and Study 1, participants from working-class (vs. middle-class) contexts are less likely to perceive voting as a duty (M_wk = 4.854, M_md = 5.663, M_diff = $-$0.808, 95% [$-$0.988, $-$0.628], P$<$ 0.001). Supporting the universal benefits account, we find a significant effect of experimental condition on perceived duty, F(1, 731) $=$ 4.512, P$=$ 0.034, ${\eta }_p^2$$=$ 0.006, suggesting that, regardless of participant social class context, individuals in the interdependent condition expressed greater perceived duty to vote (M_int = 5.349) relative to those in the independent condition (M_ind = 5.154, M_diff = 0.195, 95% [0.015, 0.375], P$=$ 0.034; Fig. 2, Top). Mirroring Study 2, our results do not support the unique working-class benefits account: We find no significant interaction between experimental condition and social class, F(1, 731) $=$ 0.568, P$=$ 0.451, ${\eta }_p^2$$=$ 0.001.

Fig. 2.

Effect of interdependent (vs. independent) condition on perceived voting duty (Study 3). Note: Error bars reflect SE. Top panel: full sample. Bottom Left panel: working-class sample. Bottom Right panel: middle-class sample. Duty composite (Avg) is a composite measure of duty, importance, and interest.

To further test the universal benefits account, we conduct a mediation analysis controlling for participant social class. This analysis suggests neither a significant direct effect of the interdependent condition on political engagement intentions (b $=$ 0.005, SE $=$ 0.071, 95% [$-$0.133, 0.144], P$=$ 0.939), nor a significant total effect (b$=$ 0.115, SE $=$ 0.088, 95% [$-$0.058, 0.287], P$=$ 0.192). We observe a significant positive indirect effect of the interdependent (vs. independent) condition on political engagement intentions via perceived voting duty, b $=$ 0.109, SE $=$ 0.053, 95% [0.006, 0.213], P$=$ 0.039.^**

While we find no significant interaction between experimental condition and social class in predicting perceived duty to vote, in light of our preregistration, we analyze both subgroups separately. For participants from working-class contexts (n$=$ 421), we observe a significant effect of condition, F(1, 332) $=$ 5.928, P $=$ 0.015, ${\eta }_p^2$$=$ 0.02, suggesting that individuals from working-class contexts were significantly more likely to perceive voting as a duty in the interdependent (M_int = 5.015) vs. independent condition (M_ind = 4.685, M_diff = 0.330, 95% [0.063, 0.596], P $=$ 0.015, Fig. 2, Bottom Left). For participants from middle-class contexts (n$=$ 404), we find no significant effect of condition, F(1, 319) $=$ 1.744, P$=$ 0.188, ${\eta }_p^2$$=$ 0.005. While those from middle-class contexts were also more likely to perceive voting as a duty in the interdependent (vs. independent) condition, this difference was not statistically significant (M_int = 5.745, M_ind = 5.579, M_diff = 0.166, 95% [$-$0.081, 0.413], P$=$ 0.188; Fig. 2, Bottom Right).

We preregistered a moderated mediation analysis with experimental condition (0 $=$ independent, 1 $=$ interdependent) as the predictor, perceived duty to vote as the mediator, and political intentions as the outcome with participant social class (0 $=$ middle-class, 1 $=$ working-class) as a moderator of all three paths. For individuals from working-class contexts, we find a small but statistically significant positive indirect effect of the interdependent (vs. independent) condition on political engagement intentions via perceived voting duty, b$=$ 0.147, SE $=$ 0.075, 95% [0.000, 0.295], P$=$ 0.05. For individuals from middle-class contexts, in contrast, we find no significant indirect effect, b$=$ 0.073, SE $=$ 0.076, 95% [$-$0.076, 0.223], P$=$ 0.337.

In sum, Study 3 further supports the universal benefits account (H2): When provided the opportunity to reflect on interdependence (vs. independence), individuals from working-class and middle-class contexts alike are more likely to perceive voting as a duty, which, in turn, is associated with more interest in engaging politically. Like Study 2, the results of Study 3 do not support the cultural match account (H1) as interdependence seems to benefit everyone. And, like Study 2, while subgroup analyses do not support the notion that interdependence is uniquely beneficial to individuals from working-class contexts, these analyses do tentatively suggest more substantial effects of interdependence (vs. independence) on perceived voting duty for individuals from working-class (vs. middle-class) contexts.^††

General Discussion

A functioning representative democracy requires citizens to be politically engaged, and while many Americans vote, a substantial portion do not. Moreover, particular groups, like those from working-class contexts, are especially underrepresented in United States politics (12). Whereas prior work has pointed to structural and individual factors that shape voter turnout, the present work adopts a sociocultural lens to shed light on an additional contributor: People’s construal of the duty to vote. First, we document rates of perceived voting duty, both across social class contexts and in the U.S. population overall, demonstrating that perceived duty to vote is meaningfully associated with voter turnout and helps explain class-based gaps in voting (Study 1). We further suggest that intervening on people’s construal of the duty to vote might serve as an effective pathway to boost perceived duty and political interest. Encouragingly, we find that interdependent construals of voting duty (i.e., voting as a duty to others)—whether spontaneous or experimentally induced—are associated with heightened perceived duty to vote. Specifically, we find that regardless of people’s social class context, construing voting as an interdependent duty is positively associated with perceived duty to vote (Study 2). Beyond correlational evidence, we find that experimentally reflecting on interdependence (vs. independence) increases perceived voting duty for all (Study 3).

Practically, the present work points to the potential of construal interventions to boost perceived duty and, in turn, motivate political engagement as a relatively costless and easily scalable complement to existing structural and individual interventions. Political campaigns and institutions could adopt this approach by simply emphasizing an interdependent (vs. independent) construal of voting in their messaging. A construal-based intervention requires only a change in the framing around why there is a duty to vote. Interventions to shift construal have effectively motivated behavior in other domains (33, 34, 47), and the present work joins and extends this work by demonstrating the potential for construal-based interventions in the political domain. Moreover, this research adds to a body of evidence by documenting how construal interventions may be effective not only for longer-term experiences like higher education (30, 56), but also for one-off behaviors (e.g., voting).

The present work is not without its limitations, and we see many fruitful avenues for future research. Most notably, our experiment does not measure actual voting. Thus, we can only conclude that reflecting on interdependence prompts individuals to endorse greater overall interest in voting, which predicts subsequent behavioral intentions to vote, but we cannot be sure that it affects their actual voting likelihood. Though evidence is mixed (57), there are several reasons to expect connections between intentions to vote and voting. First, in Study 1, we observe a significant, positive correlation between self-reported voting intentions and validated voting (r$=$ 0.29, P$<$ 0.001). Second, influential theories of behavior change consider intentions to be direct precursors to behavior (58). More specifically, empirical examinations of the relationships between voting intentions, self-reported voting, and validated voting behavior demonstrate that the three are robustly correlated with one another (59). Nevertheless, an important endeavor for future research is to examine whether having people reflect on interdependence (vs. independence) prompts changes to actual voting likelihood.

In addition, while our evidence suggests that interdependence boosts perceived voting duty, it is unclear which elements of interdependence are at work. The efficacy of different facets of interdependence (vs. independence) may depend on the particular domain in question; for example, experiencing the self as interdependent involves connection to others (vs. self-containment), whereas defining the self as interdependent involves similarity to (vs. differences from) others (60). Because our Study 3 manipulation emphasized multiple aspects of interdependence (e.g., Relationships with Friends and Family, Listening to Others, Working Together with Others, Belonging to a Group, etc.), we are unable to assess precisely which facets increased perceived voting duty. Prior work finding universal benefits of interdependence tends to focus on messages emphasizing “working together” (33–35), tapping the “connection to others” element of interdependence. In contrast, other work finding unique benefits of interdependence for those with interdependent models of self (42) tends to emphasize interdependence as adjusting to others, perhaps tapping the “receptiveness to influence” facet. In sum, certain facets of interdependence may be more or less aligned with the relatively independent selves of those from middle-class contexts. Future work would benefit from isolating the aspects of interdependence that are universally beneficial vs. uniquely beneficial to those from working-class contexts, particularly in the political domain.

Moreover, results from Study 3 suggest that reflecting on interdependence (vs. independence) has a significant indirect effect on political engagement intentions via our proposed mechanism of perceived duty, but we observe no significant total effect of interdependence on political intentions. Initial understandings of mediation assumed a significant total effect as a necessary precondition for testing mediation (61). However, more recent work finds that significant indirect effects can occur in the absence of significant total effects (62). This work therefore proposes a shift toward assessing the magnitude and significance of indirect effects. One possibility for why we observe no significant total effect is that our intervention focused on interdependence in general rather than applied specifically to the political context. It is also possible that our Study 3 intervention did not have an effect on political intentions and our observed indirect effect is an artifact of unmeasured confounds. Further, while we suggest potential mechanisms underlying the observed effects in Study 3, our data do not allow us to distinguish between our proposed accounts. Future work would benefit from tests of these mechanisms using interventions that explicitly reframe voting as an interdependent duty to examine whether manipulations applied directly to the political context have significant direct and total effects on political engagement intentions (see SI Appendix, Study 2 in SI Appendix, section 6 for an initial exploration of this possibility).

Here, we examine voting through a sociocultural, rather than individual or structural, lens. Where an individual model of behavior attributes differences to people’s personal characteristics and a structural model implies differences stem from environmental factors, a sociocultural perspective suggests that individual and structural forces mutually constitute one another (27). While we attempt to control for structural characteristics influencing voter turnout, we recognize that structural barriers are likely to be increasingly critical in predicting voting behavior (63). Future work should investigate whether messages reframing voting as an interdependent (vs. independent) duty might be used to buffer people from the increased difficulty of voting. In addition, while we control for racial group membership in our analyses, our samples are primarily White individuals, making it important for future research to examine how social class and race intersect to predict turnout (7).

Testing strategies in controlled, experimental contexts lays critical groundwork for future field studies. Indeed, the controlled environment of the experiment allows for precision and an internally valid demonstration of intervention efficacy (64), enabling us to isolate the effectiveness of intervening on sociocultural factors. Furthermore, the online setting provides access to individuals from working-class contexts, a hard-to-reach population often overlooked in voter turnout studies. Nevertheless, researchers might test these strategies in the field by partnering with a voter turnout organization and conducting a large-scale construal intervention to examine whether reframing voting as an interdependent (vs. independent) duty increases voter turnout.

Finally, future work would benefit from testing construal-based interventions in other critical domains (e.g., health behaviors, environmental behaviors, etc.) and specifically, should examine whether the benefits of interdependent (vs. independent) construals of behavior can generalize beyond the political domain. For example, how might construing vaccination as a duty to one’s community (rather than as a duty to self) affect vaccine uptake? How might reframing healthy eating and exercise as a duty to one’s children shape health outcomes? Importantly, the work here suggests that we might boost political engagement by simply changing how we—collectively—make sense of the duty to vote.

Materials and Methods

This research complies with ethical regulations. Study 1 was conducted on publicly available data. The design of Studies 2 and 3 was approved by Emory University’s IRB (00002707). All participants provided informed consent. Studies 2 and 3 were preregistered (see SI Appendix, section 1 and below for links). Study materials (data, syntax, stimuli) have been deposited on https://osf.io/4qnp6/ (65). All data were collected using Prolific Academic unless otherwise specified. For working-class samples, we targeted Prolific participants without a 4-y college degree. For middle-class samples, we targeted participants with at least a 4-y college degree.

Study 1.

Data for Study 1 came from 10,185 participants who responded to the ANES 2012 or 2016 Time Series surveys (M_age = 48.5, SD $=$ 18.5; 47.4% male, 52.0% female, 0.1% other; 6,547 White, 1,418 Black, 240 Asian/Asian American, 61 Native American, 1,459 Hispanic/Latino, 460 N/A; 34.8% middle-class; 65.2% working-class). Since 1948, ANES have conducted surveys of political opinion and behavior in the United States targeting U.S. citizens ages 18 or older. Both the 2012 and 2016 ANES surveys contain two independently drawn probability samples, one sample interviewed face-to-face and one interviewed through the internet (see SI Appendix, section 2 for links to ANES documentation). In both years, respondents are first interviewed during the 2 mo prior to the November election (pre-election survey) and are reinterviewed in the 2 mo after the November election (postelection survey). Because we use the full sample of participants (i.e., both modes of interview and variables from pre- and postelection surveys), our analyses use the weight_full variable for ANES 2012 and the V160102 variable for ANES 2016.

Participants in both years were asked during the pre-election survey whether they consider voting to be a duty, a choice, or neither (2012: variables “preswin_dutychoice” or “preswin_choiceduty”; 2016: variables “V161150a” or “V161150b”; see SI Appendix, section 2 for exact question wording). We recoded this variable prior to analysis such that 1 $=$ duty and 0 $=$ neither or choice. Participants also indicated their highest level of education (measured in the pre-election survey). Both years included a validated postelection measure of whether participants voted in that year’s general election.

We controlled for participant political conservatism (1 $=$ extremely liberal, 7 $=$ extremely conservative), gender (male, female, other), race (White, Black, Asian, Native American, and Hispanic), age, income group, and attention to politics (recoded such that higher scores reflected greater attention to politics; all demographic variables were measured in the pre-election survey). Both years of data included information about the state in which respondents resided. As a proxy for the relative difficulty of voting, we used the Cost of Voting Index (66), scored based on a state’s voter registration deadline, voter registration restrictions, registration drive restrictions, preregistration laws, voting inconvenience, voter ID laws, and poll hours. Lower scores reflect greater ease of voting. For example, Oregon had the lowest score on this index in 2016: Oregon residents were automatically registered to vote and all registered voters were sent a ballot by mail. In contrast, higher scores reflect greater difficulty of voting and are associated with decreased voter turnout. Texas, for example, had a high score on this index in 2016 as residents were required to register in person at least 30 d prior to Election Day.

Study 2.

Study 2 was preregistered (https://aspredicted.org/7VFRDJ). We collected data from 450 participants, aiming for a final sample of 400 participants after exclusions. We sampled from equal numbers of participants from middle- and working-class contexts. In total, 462 participants completed the study and 431 passed our preregistered exclusions (M_age = 40.54, SD $=$ 13.74; 49.2% female, 48.5% male, 1.6% nonbinary, 0.7% transgender; 36 Asian, 47 Black, 25 Hispanic/Latino, 4 Middle Eastern, 7 Native American, 343 White, 3 other; 49.4% middle-class, 50.6% working-class). As preregistered, we excluded participants who missed an attention check as well as participants who indicated using ChatGPT to write their responses.

Participants read: “In the United States, people often talk about how there is a duty to vote. We are interested in learning more about why YOU personally might believe there is a duty to vote. What reasons do you think are important to consider? There are no right or wrong answers, we are interested in your ideas and opinions. Please write at least 2 to 3 ideas or reasons that come to mind to complete the statement: I personally believe there is a duty to vote because...” Following this prompt, participants were given a second prompt which asked them, “Why do you think these reasons are important? Please describe in a few sentences.” Using LIWC, we coded participant responses using the validated independence-interdependence dictionary (54). The independent dictionary encompasses themes relating to individual interest and achievement, while the interdependent dictionary includes themes pertaining to social relationships (i.e., family, belonging, support). Using this dictionary, LIWC provides the percent of Interdependent and Independent words in participant responses.

After responding to the prompts, participants indicated the extent to which they believe voting is a duty, how important they believe voting to be, and how interested they were in learning more about upcoming elections (all three items measured on a scale from 1 $=$ not at all to 7 $=$ very). We computed a composite measure which averaged all three items ($\alpha$$=$ 0.89).

Participants saw a list of political activities (items taken from ANES) and indicated their likelihood (1 $=$ not at all, 7 $=$ extremely) of engaging in each in the next 2 y. We averaged participant responses to all items as a measure of political intentions ($\alpha$$=$ 0.89). The activities asked about were the following: 1) Attend political meetings, rallies, speeches, or fundraisers, 2) Wear a campaign button, put a campaign sticker on your car, or place a sign in your window or lawn, 3) Contribute money to an individual campaign, 4) Contribute money to a political party, 5) Discuss politics with family or friends, 6) Get into a political argument, 7) Join a protest, march, rally, or demonstration, 8) Sign a petition about a political issue, 9) Post online about a political issue, 10) Contact or try to contact a member of the US Congress, 11) Work with other people to deal with issues facing your community, 12) Vote in a primary election, and 13) Vote in a general election.

Our analyses control for participant age, gender, race (Asian, Black, Latino, Middle Eastern, Native American, White, other), income group, and party affiliation (Democrat, Republican, Independent, other). In addition to being asked about their political engagement intentions, participants indicated whether they participated in any of the aforementioned activities in the last 12 mo. We summed the number of activities participants indicated having participated in as a measure of past engagement. Our preregistration indicated that our primary analysis would examine the association between interdependent (vs. independent) language and perceived voting duty for participants from working-class contexts. We preregistered that we would conduct these analyses both with and without controls (i.e., age, gender, race, income, political affiliation, and past political behavior). Our preregistration noted we would control for seriousness (i.e., “how seriously did you take the survey?”), however, due to a survey error, participants were not asked this question. We preregistered three ways of examining interdependent (vs. independent) language use. For ease of presentation, we focus on one operationalization here; however, see SI Appendix, section 3 for similar results using the additional metrics (including exploratory analyses using ChatGPT to determine the relative interdependence of participant responses). Specifically, we focus on the relative difference between interdependent and independent language use, computed by subtracting the percentage of independent language from the percentage interdependent language of the response.

Study 3.

Study 3 was preregistered (https://aspredicted.org/7X1VGQ). We collected data from 850 participants, sampling from equal numbers of those from middle- and working-class contexts. 851 participants completed the study and 825 passed both attention checks (M_age = 40.96, SD $=$ 13.17; 56.2% male, 41.2% female, 2.6% other; 52 Asian, 68 Black, 56 Hispanic/Latino, 6 Middle Eastern, 15 Native American, 689 White, 7 other; 49.0% middle-class, 51.0% working-class).

Participants were randomly assigned to an interdependent or independent condition. In both, participants read: “In this writing task, you will be answering several questions about your ideas, your beliefs, and your life.” Participants saw eight values (six independent or interdependent, two neutral) and were asked to select three values most important to them. In the interdependent condition, participants saw: Relationships with Friends, Relationships with Family, Being Loyal to Others, Belonging to a Group, Listening to Others, Working Together with Others. Participants in the independent condition, in contrast, saw: Independence, Valuing Personal Accomplishments, Learning New Things, Developing My Interests, Being Unique, Paving My Own Path. Participants in both conditions saw two neutral values: Seriousness and Calmness.

After selecting their most important values, participants wrote a few sentences about why these values are important to them. After completing the writing task, participants indicated their agreement (1 $=$ strongly disagree, 7 $=$ strongly agree) with: “In general, I try to live up to these values,” “These values are an important part of who I am,” and “I care about these values.” We computed an average and as preregistered, excluded participants who scored below the midpoint. Perceived voting duty ($\alpha$$=$ 0.90) and political intentions ($\alpha$$=$ 0.90) were measured identically to Study 2. Our controls were identical to Study 2, with the addition of a measure of how seriously participants took the survey (participants were asked, “How seriously did you take this survey?” and responded on a scale from 1 $=$ not at all to 7 $=$ very). We excluded 23 participants who missed either of two attention checks in the survey and 3 participants whose values importance scores were below the midpoint.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

Anonymized data have been deposited in Open Science Framework (https://osf.io/4qnp6/) (65).