“The personal is political” was a popular mantra that emerged in the 1960s as a rallying cry of women’s movements. This saying underscored that the challenges, dilemmas, and disruptions of life were not simply private affairs but were rooted in political choices and processes. During the 1960s and 70s, connecting the personal to the political was a strategy for spurring collective action (Combahee River Collective 1979; Nelson, 2003). Since then, the popular saliency of “the personal” has ebbed and flowed but the basic logic holds: personal life can be profoundly political. Personal crises like unemployment, eviction, and financial distress are products of a changing political economy that has multiplied risk for those with fewer resources, deepened economic inequality, and left more Americans vulnerable to destabilizing financial loss. In this paper, we assess the political consequences of such personal crises.

Economic risk and instability are staples of American life (Hacker, 2006; Western et al., 2012). Particularly since the 1970s, national crises around employment, housing, and healthcare have consistently generated interconnected and cumulative negative life events: job loss, disrupted health insurance, home foreclosure, and more. COVID-19 brought these realities into sharp relief. As the U.S. economy faltered under the weight of the pandemic, tens of millions of Americans faced resultant personal crises that were intensified by surging prices for gas, food, and rent as inflation hit a four-decade high.

We investigate the politics of personal crisis in this paper by exploring how disruptive life events affect political participation. Public opinion research demonstrates the importance of economic experiences for policy attitudes (Hacker et al., 2013), while scholarship on participation typically focuses on the effects of economic crises like recessions (Bartels, 2013) or specific life events like foreclosure (Hall et al., 2021; Shah and Wichowsky 2019), marriage (Stoker and Jennings 1995), childbirth (Kam, Kirshbaum and Chojnacki 2023), or unemployment (Burden and Wichowsky 2014). We extend this research by examining a broader scope of life events involving work, finances, housing, health, and more. Despite their heterogeneity, such events are the product of larger political processes that structure the distribution of economic risk; they therefore have implications for democratic participation (Burns et al., 2001). We contextualize our research in relation to the American economy, and we put our work in conversation with several theories of political participation to cull insights on how life disruptions can hinder or spur engagement.

We analyze data from the 2018 and 2020 Cooperative Election Study and the Democracy Fund Views of the Electorate Research survey to demonstrate that personal crises dampen voter turnout. Our analyses indicate that personal crises have heterogenous effects on political activity. As crises accumulate, voting declines for many citizens while participation in other political acts increase for a few citizens. These findings are consistent across elections before and during the COVID-19 pandemic for crises that are not related to the pandemic. Pandemic-related crises (e.g., contracting COVID-19 or losing one’s job because of the pandemic) appear to spur all forms of participation, perhaps because of the political salience of the disease. This research underscores the case for more comprehensive attention to the participatory implications of personal crises engendered by a precarious political economy.

Charting the U.S. Crisis Economy

The COVID-19 pandemic disrupted the health and prosperity of the American people more than any event in living memory. Yet, instability and crisis are nothing new in American life. Recent breakdowns in the housing, labor, and healthcare markets widely imperil economic security. As Jacob Hacker wrote in the New York Times before the pandemic, “…many Americans feel insecure. They may be doing well at the moment, but they fear that, however high they are on the economic ladder, a single step or bad event could cause them to slip” (Hacker, 2019). We contextualize our study of how “a single step or bad event” dampens political participation by first discussing three pivotal contemporary crises.

The housing crisis of the 2000s was catalyzed by diminished affordable housing and skyrocketing foreclosures. The share of high debt/low equity loans rose sharply in the 1990s. When home prices began to drop in 2006, these high leverage loans defaulted at much higher rates, driving foreclosures (Corbae and Quintin 2015). The rental housing market was similarly dire. Between 2000 and 2010, inflation-adjusted median asking rent increased between 21% and 37% while median incomes rose by only a fraction as much (Desmond, 2018). The combination of high leverage loans and increasing housing costs spurred the Great Recession, heightening economic instability for many Americans. Today, affordable housing continues to plague many Americans. As Pew Research Center noted in January 2022, “Prospective homebuyers and renters across the United States have seen prices surge and supply plummet during the coronavirus pandemic” (Schaeffer, 2022).

Though rooted in the housing crisis, the Great Recession had severe labor market repercussions. The unemployment rate jumped from 5 to 9.5% by the end of the recession, making the employment decline associated with the Great Recession greater than that of any recession since 1980 (Bureau of Labor Statistics, 2012). The sharp drop in employment across the country was due in part to a spike in the number of mass layoffs, which occur when at least 50 initial claims for unemployment insurance are filed against an establishment during a consecutive 5-week period. (Bureau of Labor Statistics, 2012). Mass layoffs peaked in February 2009 when “employers took 3,059 mass layoff actions” affecting 326,392 workers (Bureau of Labor Statistics, 2012). The specter of labor market instability extended beyond the Great Recession—even as standard unemployment returned to pre-recession levels, long-term unemployment and involuntary part-time employment remained high (Bureau of Labor Statistics, 2018).

Most recently, the Covid-19 pandemic called attention to how easily the labor market could falter. In April 2020 the unemployment rate spiked to 14.8%, the highest rate observed since the U.S. began collecting official unemployment data in 1948 (Congressional Research Service 2021). For well over a year after that peak, the unemployment rate remained higher than pre-pandemic levels. Black and Latina women, younger workers, and people in employed in the service sector experienced especially high levels of job loss (Brower and Michener 2021; Congressional Research Service 2021).

Even as the labor market rebounds from the damage of the pandemic, those who find work in the modern economy do not benefit from the same stability and security that many of their parents did. Jobs that pay well and offer key non-wage benefits like health insurance and retirement coverage have grown increasingly scarce over the past several decades, even as the average American worker has become both older and more skilled (Schmitt and Jones 2012). American employment is increasingly “polarized,” with middle-wage jobs losing ground to low-wage jobs, and much of the growth in the latter taking place in service rather than manufacturing industries (Dwyer and Wright 2019). Meanwhile, the growth of non-standard employment and the so-called “gig economy” creates jobs with precarious work schedules, financial insecurity, and distrust among workers (Lambert, Henley, and Kim 2019). In short, work in the modern economy no longer guarantees financial security.

Making for a trifecta of crises, long standing predicaments around affordable healthcare came to a head just as the American economy faced breakdowns in the housing and employment sectors. Prior to the Affordable Care Act (ACA), the uninsured rate was as high as 18%, many Americans delayed medical care due to costs (Kenney et al., 2012), and medical debt ballooned (Doty et al., 2008). Like the housing and employment crises, problems with healthcare access and affordability created precarity. The ACA reduced the uninsured rate and improved access to care (Glied et al., 2017), but these gains eroded during the Trump presidency (Galewitz, 2019) as insurance premiums and deductibles grew more quickly than wages (Claxton et al., 2018) and medical bankruptcy remained a serious threat (Himmelstein et al., 2019).

The Political Impact of Personal Crises

These national crises highlight how the political economy spurs personal crises. But do personal crises in turn shape the way citizens engage with politics? We argue that personal crises are generally demobilizing but may spur political participation in some circumstances.

How Personal Crises Can Be Demobilizing

Numerous theories and empirical literatures point to economic and psychosocial explanations for why personal crises might be demobilizing. The data and analyses in this paper do not allow us to identify one or the other of these mechanisms as the driver of our results, but we nevertheless outline a range of existing theories that are consistent with the patterns we uncover between personal crises and political participation.

A robust literature in behavioral economics demonstrates the wide-ranging consequences of scarcity, and this logic can be applied to the resource model of political participation to understand the impact of personal crises. The resource model holds that participation requires resources, such as time, civic skills, and money, and so “the presence or absence of resources contributes substantially to individual differences in political participation” (Verba et al., 1995, 274). Since the resources for participation are finite and in demand, individuals must evaluate the opportunity costs of how and when they should be deployed. Finiteness is the limit or bound to the supply of a resource. There are only 24 h in a day and only so many dollars in a bank account. Once these hours or dollars have been spent on one activity they cannot be used again for another activity. Civic skills are less finite because they are not lost once used. The ability to write a letter to a representative can be deployed repeatedly without depleting its supply. Nevertheless, the deployment of civic skills depends on mental bandwidth—an individual’s attentional and cognitive functioning (Mani et al. 2013; Mullainathan and Shafir 2013; Schilbach et al., 2016)—which is finite. Few could routinely write several letters a day even if they found the time and supplies to do so. Importantly, the resources for political participation are also required of other activities. Money is required to make a political donation, but it is also required to purchase food and clothes, pay for shelter and transportation, and participate in leisure activities. Time and mental bandwidth are likewise in demand. Feeding oneself requires both time and attention in addition to money—planning meals, making a grocery list, purchasing goods, and then preparing the food for consumption are not costless activities.

That resources are finite and in demand means they are scarce, and this scarcity forces people to make decisions about how to allocate their limited supply of resources. While decisions over how to use resources will be somewhat idiosyncratic, individuals will generally give basic needs primacy over higher-order needs (Maslow, 1943). Basic needs are shelter, food, clothing, utilities, and sanitation, intermediate needs are personal security, employment, health, property, family, and friendship, and higher order needs are status, leisure, and political participation. Consider an individual who faces a home foreclosure as an illustrative example. The process of foreclosure will increase the demand for resources—money, time, and mental bandwidth to relocate; mental bandwidth and time to remain organized in the process of moving; time and energy to deal with emotional turmoil; and so forth. Resources expended for these purposes are no longer available for political participation. Simply put, the opportunity costs of participation are greater when basic needs lead to a heightened demand for resources, such as the demands created by personal crises.

There are also psychosocial reasons to think personal crises may be demobilizing. An abundance of research finds that the experience of stress, whether daily hassles or major traumas, can be deleterious to cognitive functioning, mental well-being, and physical health, especially as it intensifies, accumulates, and persists (Hammen, 2005; Schneiderman et al., 2005; Sandi, 2013). One review of the literature on stress and cognition concludes, “the general view that emerges is that exposure to high-to-very-high stress acutely (whether elicited by the cognitive task or experienced before being trained or tested in the task) or chronically impairs performance on explicit memory tasks that require complex, flexible reasoning while improving performance on implicit memory tasks, in simple declarative memories and in well-rehearsed tasks” (Sandi, 2013, p. 255). This is important because these factors facilitate political participation (Denny and Doyle 2008; Landwehr and Ojeda 2021; Pacheco and Fletcher 2015), potentially because cognitive functioning is a resource for participation. To the extent these resources are diminished due to personal crises, participation would be expected to decline.

Personal crises, and their impact on cognition, mental well-being, and physical health, may also impair the motivation to participate. Depression is one indicator of poor mental well-being and has been found to reduce political interest and internal political efficacy, factors that motivate political participation (Landwehr and Ojeda 2021; Ojeda and Slaughter 2019). Similarly, poor physical health has been linked to lower political trust (Mattila, 2020), while reduced cognitive functioning has been found to diminish political trust (Hooghe et al. 2012) and political interest (Denny and Doyle 2008). In short, there are good reasons to think that personal crises may be demobilizing because they reduce our motivation and ability to participate.

When Personal Crises are Mobilizing

Just as there are economic and psychological reasons to think personal crises are demobilizing, there are economic and psychological reasons to think they may sometimes mobilize participation. Regarding the economics of personal crises, it is possible that extreme crises spur participation insofar as people see politics as the only route to solving their problem and ensuring their basic needs are met. However, such perceptions are likely only in cases where there is a direct relationship between the participatory act and the crisis. For instance, the threat of foreclosure may prompt a protest outside of the bank initiating the foreclosure or attendance at a local meeting about foreclosures in the neighborhood. It may also alter attitudes towards government regulation of banks or government programs designed to help homeowners. Absent such direction connections, the reality of finite resources in the context of a hierarchy of needs suggests crises will likely decrease participation. This is indeed borne out in empirical research (Estrada-Correa and Johnson 2012; Hall et al., 2021; Shah and Wichowsky 2019).

This dynamic is particularly germane to assessing how personal crises might affect different forms of political participation differently. While voting is a central component of democracy, it is also a generalized form of participation that does not ordinarily pertain to specific material needs, and so can be viewed as a higher-order activity. In contrast, other forms of participation (e.g., contacting a local official, attending a meeting, protesting) can sometimes relate directly to basic needs and can thus provide a greater possibility of relief. We therefore expect crises to be demobilizing with respect to voting. We are less certain about how they might affect other participatory acts.

It is also possible that some personal crises place a demand on one resource while simultaneously increasing the availability of another resource. Unemployment, for instance, reduces money but increases free time (Aguiar et al. 2011). These “mixed” demands might be expected to have a diminished impact on political participation to the extent that the resources for participation are fungible (e.g., free time can be substituted for money) (Burden and Wichowsky 2014; Cebula, 2017).

From a psychosocial perspective, there are also good reasons to think personal crises will sometimes be mobilizing. In addition to diminishing cognitive bandwidth, mental well-being, and physical health, personal crises can evoke negative emotions like anger and anxiety (Spielberger et al. 1984). It is well-documented that anger, and to a lesser extent anxiety, has a mobilizing effect (Phoenix, 2019; Valentino et al., 2011), creating the possibility that personal crises spur participation some of the time. However, for emotions to be systematically rather than merely incidentally mobilizing, they must be directed at government and politics. This seems possible in circumstances where personal crises are clearly connected to salient issues for which government is viewed as bearing some causal and/or treatment responsibility (Arcenaux 2003).

To summarize, based on existing economic and psychosocial explanations of political behavior, we argue that personal crises are likely to dampen participation by squeezing scarce resources and sapping motivation, except when the political act offers the potential for relief or when there is high political saliency, clarity of responsibility or perceived responsibility. This picture of disruptive events as politically demobilizing is in tension with notions of democratic accountability—the idea that adverse societal outcomes mobilize citizens to punish incumbent politicians (Healy and Malhotra 2013). Our argument is not that catalyzing political action amid crisis is impossible, but rather that political action generally becomes more difficult when resources and motivation are depleted. Again, this may explain why research finds that experiencing foreclosure during the Great Recession dampened turnout (Hall et al., 2021)—although the salience of the Great Recession was high, perceptions about who was responsible were muddled (Parker-Stephen, 2013) and motivated reasoning prevailed in the minds of voters (Bisgaard, 2015).

In any case, we need not look far for evidence that crises dampen turnout. Studies find that a wide range of crises reduce turnout, including teen pregnancy and dropping out of school (Pacheco & Plutzer 2007), poor health (Burden et al., 2017; Ojeda and Pacheco 2017), the death of a spouse (Hobbs, Christakis, and Fowler 2014), divorce (Dehdari et al., 2022), the loss of Medicaid (Haselswerdt and Michener, 2019), unemployment (Burden and Wichowsky 2014), foreclosure (Hall et al., 2021; Shah and Wichowsky 2019), acute financial hardship (Schaub, 2021), loss of income (Schafer et al., 2022), and the death of a loved one from opioid overdose (Kaufman and Hersh 2020). The analyses offered below build on these studies but take a broader approach by considering patterns across a wide range of crises, varying contexts of crises (before and during the COVID-19 pandemic), and different forms of political participation.

Methodology

We test our argument using data from the Cooperative Election Study (CES; formerly the Cooperative Congressional Election Study) Common Content and Modules in 2018 and 2020 as well as the Democracy Fund Views of the Electorate Research (VOTER) Survey. The CES is an annual survey administered by YouGov and fielded to a nationally representative sample of 60,000 + respondents, with the opportunity for teams of researchers to field original questions to 1,000-respondent modules. We draw on questions from the 2018 and 2020 CES Common Content—supplemented with a battery of original questions on team modules in both years—that measure personal crises, such as loss of health insurance, loss of home, eviction, loss of eligibility for a government program, loss of eligibility for a tax break, trouble affording medical expenses, substantial increases in housing costs, and suspension or revocation of driver’s license.

The VOTER Survey is an ongoing panel study of American adults administered by the Democracy Fund’s Voter Study Group in cooperation with YouGov. The May 2018 wave of the survey include a series of questions on financial difficulties encountered in the twelve months preceding the survey, including the loss of job (self or spouse), problems paying mortgage or rent, problems paying a student loan, problems paying a car loan, problems paying off a credit card, and drop in household income. Table 1 summarizes the key features of each study. Appendix A elaborates on the data collection process, while Appendix B describes the measurement of all variables used in the analysis.

Table 1 Key features of the datasets
Full size table

The primary dependent variable in all studies is validated voter turnout. Information about the voter-file matching procedure is reported in Appendix B. We also estimate models using self-reported non-voting forms of participation for the CES studies, including attending a political meeting, putting up a political sign, working for a candidate or campaign, attending a protest, contacting a public official, and donating money.

We draw on stress research to inform the measurement of our independent variable, which we refer to as personal crises. We discuss three issues here: operationalization (i.e., how to define personal crises), measurement (e.g., how to generate data on personal crises), and aggregation (e.g., how to create a single variable out of data on many personal crises). Our focus here is explaining how we arrived at our independent variable; in the discussion, we describe other measurement approaches that could be used in future research.

The first issue is how to operationalize personal crises. Our focus is on “objective experiences that disrupt or threaten to disrupt an individual’s usual activities” (Dohrenwend & Dohrenwend, 1969, p. 133) and “cause a substantial readjustment in that person’s behavior” (Thoits, 1982, p. 342), such as divorce or losing a job.Footnote 1 Stress research consistently finds that the experience of personal crises leads to psychological distress and physical illness (Sheldon et al. 2019). Meanwhile, as we noted earlier, studies in political science have examined how particular personal crises, including traumas, impair participation (e.g., Marsh, 2022; Hobbs et al., 2014). This makes personal crises a good starting point for thinking about how the cumulative stress burden affects political participation.

The development of personal crises checklists, such as the Social Readjustment Rating Scale (Holmes and Rahe 1967) or the updated and modified Stressful Life Events Scale (Hobson et al., 1998), have assisted scholars in identifying relevant, discrete, independent, and unbiased events across domains such as finances, family, health, and mortality. We included an abbreviated set of 13 personal crises in the 2018 CES Module and then expanded to 45 personal crises in the 2020 CES Module, which is nearly the full complement from the Stressful Life Events Scale. The CES Common Content and VOTER Survey, which we did not design, include a more limited sets of items, with the latter focused exclusively on financial difficulties.

The second issue is how to measure personal crises. The most common approach, and the one we use here, is to focus on incidence. Incidence refers to whether a person experienced a particular crisis, and it is shown to be predictive of negative outcomes (Monroe, 2008). We measure incidence by asking respondents whether they experienced each crisis in the past year. Recognizing that not all crises are the same, scholars sometimes measure perceptions of frequency, length, and severity of crises (e.g., Friborg, 2019). We generally lack the necessary data to categorize crises on these dimensions without making unsupported assumptions, but we address how future research can fruitfully tap into other dimensions of personal crises in the discussion. Importantly, a measurement strategy focused only on incidence likely represents the lower bound for the impact of crises on political participation. We might expect a stronger effect were we to account for other aspects of crises.

The final consideration is aggregation. We generate a single score for each respondent by counting the incidence of personal crises. In our case, a count is more appropriate than a factor analysis (or a similar data reduction technique) because personal crises are the cause rather than the effect of stress. Factor analysis is useful when we expect indicators to be highly correlated because they are caused by the same latent variable; however, it can introduce bias when indicators (such as ours) are not strongly correlated nor expected to be (Bollen and Ting 2000). A count approximates the total stress burden over the past year, giving equal weight to each crisis. In the discussion, we consider how future research can explore alternative ways of aggregating crises, such as weighting by the severity or length of the crisis or its proximity to election day.

Altogether, the main independent variable is the total crises experienced by each respondent in the past year. However, we undertake a preliminary exploration of two different types of crises. We create separate scales in the CES 2020 for non-COVID and COVID crises to capture the unique politics surrounding the pandemic (Gadarian et al., 2022). We also present a supplementary analysis in the appendix that distinguishes between crises that more likely represent ongoing problems in a person’s life from those that more likely represent significant changes to the status quo.

At least one personal crisis is reported by 42% of respondents in the 2018 CES Common Content, 60% in the 2018 CES Module, and 34% in the VOTER study. For the 2020 CES, at least one non-COVID crisis is reported by 48% of respondents in the Common Content and 69% in the Module, while at least one COVID crisis is reported by about 61% of respondents in the Common Content and 59% in the Module. The differences between surveys are largely a function of the total number of crises asked about on the survey—the more crises that are included, the smaller the percentage of respondents reporting no personal crises. As Figure 1 shows, the percentage of respondents reporting a given number of personal crises declines as the overall number of crises surveyed increases.

Fig. 1
figure 1

Distribution of personal crises

Full size image

We estimate logistic regressions of validated voter turnout and other forms of self-reported participation on the count of personal crises. The CES 2020 models include separate variables for non-COVID and COVID crises. A negative and statistically significant coefficient would provide evidence that crises inhibit participation. All models employ survey weights and control for validated turnout in the 2016 election, age, education, income, gender, race, marital status, religious attendance, and the presence of children under 18 in the home.Footnote 2 The control variables in the CES are measured concurrently with the key dependent and independent variables, aside from validated 2016 turnout, while all control variables in the VOTER Survey are measured in 2016. Appendix B displays question wording and other measurement details, Appendix C reports descriptive statistics, and Appendix D reports the incidence of each crisis and distributions of the totals.

Results

We begin by examining the bivariate association between each personal crisis and turnout. Figure 2 displays the difference in mean turnout between respondents who did and did not experience each crisis in the CES 2020 Module, which included the most extensive battery of personal crises. The results show a consistently negative turnout gap for non-COVID crises: 33 of the 36 crises (about 92%) create a negative gap, while 19 of the 36 crises (about 53%) create a negative and statistically significant gap. No non-COVID crises exhibit a positive and statistically significant turnout gap. A different pattern emerges for COVID crises: only 4 of the 9 crises (about 44%) are negative, while only 2 of the 9 crises (about 22%) are negative and statistically significant. Unlike with non-COVID crises, 2 of the COVID crises create a positive and statistically significant gap.

Fig. 2
figure 2

Association of specific personal crises with the probability of voting

Full size image

Table 2 summarizes a series of t-tests, like the ones reported in Fig. 2, for all five studies. The results show that almost all non-COVID crises are negatively associated with turnout, and most of these associations are statistically significant. The differences in turnout for the five crises with a positive association are not statistically significant. These results are consistent with previous “single crisis” studies, but also point to a systematic pattern in which personal crises are broadly consequential to voting. The COVID crises exhibit a mixed pattern, with fewer than half of the nine measured events showing a negative association with voting in both the 2020 CES Common Content and Module. Furthermore, some of the positively associated COVID crises are statistically significant.

Table 2 Bivariate associations of specific personal crises with voting across studies
Full size table

We now turn to the impact of crisis accumulation on political participation. Table 3 shows turnout levels across weighted tercile of total crises for each study. The results for non-COVID crises once again align with our expectations: those who experienced more crises are less likely to vote. This difference is substantial—as high as 19.5% points in the 2018 CES module. For context, this disparity is roughly four-fifths the magnitude of that between non-Hispanic White and Hispanic respondents in the same survey (23.8% points). The differences shown in Table 3 are statistically significant in all four CES surveys. In the VOTER Survey, the difference (though negative and monotonic) is smaller and not statistically significant. These findings are consistent with the idea that the totality of disruptions to everyday life can be disruptive to turnout. While correlation is not causation, this pattern is noteworthy from a purely descriptive standpoint: people with unstable lives are systematically underrepresented at the ballot box. At the same time, we observe the opposite relationship for COVID-related crises, with higher turnout among those who experienced the greatest number of pandemic-related events, though the difference is only statistically significant in the 2020 CES Common Content.

Table 3 Voting turnout across terciles of personal crises
Full size table

We present the results of the multivariate models in Table 4, focusing only on the independent variables of interest; the full results are reported in Appendix E. Since these models control for validated past turnout in addition to demographic variables, they are less vulnerable to selection bias than the previous results. We observe a negative coefficient for non-COVID crises across all models. The coefficient magnitudes are consistent across models—ranging from − 0.091 to -0.109—and statistically significant in three of the five specifications, with the exceptions of the 2018 CES Module (p = 0.30) and the VOTER Survey (p = 0.12). These latter coefficients are comparable in size but not as precisely estimated. We speculate that the higher standard errors may be due to both smaller sample sizes and smaller batteries of crisis questions in those surveys. Unlike non-COVID crises, the COVID crises variable has positive and statistically significant coefficients in the two CES 2020 models. These findings support the idea that personal crises generally reduce turnout, while highly politicized ones may spur it.

Table 4 Logistic regressions of voting on personal crises
Full size table

Using the models in Table 4, we estimate that a single non-COVID crisis reduces the probability of voting by about 1% point, while a shift from 0 to the 95th percentile in crises reduces the probability of voting by as much as 10% points in the CES 2020 Module and 13% points in the CES 2020 Common Content. For context, these models predict an increase in turnout of 1.2 to 2.3% points for a one-unit increase in the ordinal education measure (e.g., moving from “high school graduate” to “some college”), which is larger in magnitude than the dampening effect of experiencing one additional non-COVID crisis, but not dramatically so. Moreover, since the racial/ethnic and socioeconomic markers of disadvantage accounted for in these models are also associated with increased incidence of crisis (see Appendix F), the coefficients reported in Table 4 may understate the dampening effect of these experiences. Indeed, such experiences may themselves constitute a partial explanation for racial, ethnic, or socioeconomic disparities in voting. Turning to COVID-related crises, a shift from 0 to 3 crises (the 95th percentile value) is associated with an increase in turnout of 4% points in the Common Content and 12% points in the Module.

We explore the possibility of heterogeneous effects in Appendix G and find that the negative relationship of non-COVID crises with voting is strongest among those who voted in 2016. The interaction of past voting with non-COVID crises is negative in all five models and statistically significant in four, including the VOTER study for which we found null effects in the original model. This suggests that crises may disrupt habitual turnout. We find no consistent evidence that effects are conditional on other covariates.

In a separate analysis of the 2020 CES Module, we distinguish between crises that could be ongoing in a respondent’s life and those that more likely represent a new significant change. Since respondents were only asked to report on the past twelve months, we have no direct measure of whether a given problem is ongoing or new, and so the coding is based on our best judgment. To illustrate this distinction, “serious trouble with romantic partner” may represent an ongoing issue, while divorce is a specific qualitative change in a person’s status. These results, and a more extensive explanation of our coding decisions, are reported in Appendix H. We find that the effect of ongoing problems approaches zero, perhaps because the past turnout control accounts for chronic effects, while recent change drives the significant and negative effect. This pattern may indicate that sudden changes or disruptions are more detrimental to voting.

We now turn our attention to how personal crises shape other forms of participation: attending a local political meeting, putting up a political sign, working for a candidate or campaign, attending a political protest, march, or demonstration, contacting a public official, and donating money to a candidate, campaign, or political organization.Footnote 3 The six participation variables scale well (α > 0.7 in all four CES datasets), and so we also create an index of participatory acts. We estimate a regression model for each dependent variable—OLS for the index and logit for the dichotomous variables—using total personal crises as the key independent variable.Footnote 4 Table 5 displays the results for the personal crises variable in each model.

Table 5 Regressions of non-voting political acts on personal crises
Full size table

The results show mostly positive and often statistically significant effects, consistent with the idea that non-voting political activities are more targeted than voting and therefore provide a potential solution to a personal crisis; in this sense, voting is best characterized as a higher-order activity that someone in desperate circumstances cannot readily afford. Other political acts may offer hope of relief, or a channel for expressing anger, and therefore be judged worthy of the resources they demand (Michener, 2019).

Importantly, these other forms of participation are much rarer than voting – in 2020, for example, validated voter turnout in the CES was 62%, while self-reported participation ranged from 3% (working for a candidate or campaign) to 23% (donating to a candidate, campaign, or organization). Furthermore, validated voter turnout is very high (sometimes well over 80%) among politically engaged citizens who participate in one or more of these other acts. While the data are limited, turnout appeared to be similarly high among people who experienced one or more crises and engaged in one or more non-voting acts. It is important to avoid the ecological fallacy that these findings suggest at first glance: it is unlikely that personal crises make any one person less likely to vote and more likely to engage in other participation. Rather, we suspect that the depressive effect on voting and the positive effects on other forms of participation play out among different subsets of citizens.

A plausible alternative explanation for these findings, however, is that experiencing crises depresses all forms of political participation for those with less propensity to participate but boosts all forms of participation for those with a higher propensity. Recall, however, that our heterogeneous effects analysis of voter turnout found negative and statistically significant interactions of crisis incidence with validated past turnout, which militates against the argument that crises increase turnout among those who are already engaged. We also looked for heterogeneous effects of crises across prior (pre-crisis) political interest in the VOTER survey. The results, which are reported in Appendix G, do not show evidence of an interactive effect (p = 0.37). These exploratory analyses suggest that there is no reason to believe that the negative effects of crisis on voter turnout are reversed for highly engaged citizens.

Discussion & Conclusion

The enduring recognition that the “personal is political” (Combahee River Collective 1979), along with important recent changes in our politics and economy are enough to warrant continued consideration of how personal crises affect political participation. In taking up this important topic, we are especially attuned to the broader political economy, the concrete realities of people’s lives, and the multifaceted theoretical logic of political participation. Empirical tests reveal that most personal crises are associated with depressed voter turnout and that turnout declines as crises accumulate. The apparent relationships between these crises and turnout approach the magnitude of well-established disparities in the turnout literature (namely the gap between non-Hispanic White and Hispanic Americans and those of different levels of educational attainment). We also find robust evidence that these associations are larger among those who voted in the previous election.

At the same time, our findings show that crises do not always and everywhere lower engagement. We identify two scenarios where crises may spur participation in a way that is consistent with classic models of political accountability. First, the likelihood of voting may increase rather than decrease when a potential voter is affected by a crisis that is highly salient and deeply political, such as the COVID-19 pandemic. Second, while crises in general may dampen the likelihood of voting, they may also make at least some citizens more likely to engage in other forms of participation, like protesting, contacting their elected officials, or volunteering for a campaign. In both scenarios, we suspect that the underlying mechanism is the same: the citizen can see participation as a means of addressing the problem that affected them. While voting is usually a very generalized form of political participation that is not tied to any one issue, that can change when a crisis is highly salient and highly politicized, as the COVID-19 pandemic has been. Other forms of participation, meanwhile, tend to be much more targeted to specific issues.

While we cannot make any dispositive claims, we do carefully and thoroughly leverage a wide range of available data to investigate the politics of life disruptions. Scholars can readily build on our findings through experimental, quasi-experimental, qualitative, and further observational approaches. Altogether, we offer a valuable contribution that advances knowledge of the politics of personal crises and provides a foundation on which future research can build.

While we situate our inquiries capaciously within both economic and psychological theories of political and social action, the analyses do not exhaust the full range of complexities and relationships worth understanding vis-à-vis the politics of personal crises. For instance, we do not comprehensively examine interactions between different kinds of crises, the distinct effects of combinations of crises, or the implications of the differential intensity of crises. It is quite likely that personal crises are not only cumulative but also interactive and interrelated. Moreover, even the same crises can be experienced at different levels of intensity by those facing them. Scholars seeking to better understand the political consequences of personal crises in relation to some of these complexities could gain traction by collecting more and better kinds of quantitative data. For example, measuring personal crises in panel data sets that follow the same people over a long period of time would provide valuable empirical leverage, while including survey questions gauging respondents’ sense of the intensity, length, and specific timing of crises would allow researchers to account for crisis heterogeneity. In addition to such survey based quantitative work, careful qualitative work can help to build knowledge of how crises are differentially perceived across stages of the life course, at varied levels of intensity, across relevant groups (e.g., people living in poverty versus middle class people), and more. Close qualitative investigation could also help scholars to chart new ways people articulate the connections between personal crises and political choices. Altogether, this topic is fertile ground for substantively important and methodologically wide-ranging research going forward.

Most immediately, our results demonstrate the value of examining the politics of personal crisis with nuance and comprehensiveness. Instead of focusing on a single category or a particular economic shock, we considered a wide and varied spectrum of challenges that emerge in the context of a precarious economy. While this research illuminates the political economy of personal crisis, continued elaboration of the mechanisms and processes structuring the connection between personal crisis and political outcomes remains an important task. As Americans cope with the devastating repercussions of a global pandemic, understanding the democratic consequences of personal crisis takes on a new and crucial importance.