The entry into risk society occurs at the moment when the hazards which are now decided and consequently produced by society undermine and/or cancel the established safety systems of the welfare state’s existing risk calculations.
—Beck (Reference Beck1999)Technological developments over the past 15 years have dramatically reshaped the way that many people earn a living in the United States and around the world. The so-called gig economy, embodied by app-based platforms like Uber and TaskRabbit, has enabled millions of workers to earn incomes outside a traditional employer-employee relationship. While the concept of freelance or contract labor is not novel, app-based platforms have made this sort of work more fluid, less predictable, and more widespread. By some estimates, about one in six Americans has earned money from an online gig platform (Anderson et al. Reference Anderson, McClain, Faverio and Gelles-Watnick2021; Katz and Krueger Reference Katz and Krueger2019; Smith Reference Smith2016; Upwork 2020). The gig economy offers considerable benefits to workers, particularly flexibility, but it also increases their economic precarity. In addition to the inherently unstable nature of this type of employment, gig workers are largely excluded from the most generous tiers of the American social safety net, which are based on traditional full-time employment: subsidized employer-provided benefits and social insurance programs like Social Security. At the same time, the legal classification of gig workers remains unsettled. While many platforms categorize them as independent contractors, countries such as Spain and Italy have taken legal steps to reclassify certain gig workers as employees, recognizing their structurally dependent position (Aloisi Reference Aloisi2021; Reference Aloisi2022). In the US, efforts to reclassify gig workers as employees also emerged under the Biden administration, but implementation of the rule was blocked under the second Trump administration (Clark Reference Clark2025). These uncertainties in employment status have important implications, not only for access to labor protections but also for how gig workers perceive their economic position and engage politically, including the types of policies they support and the forms of political participation they pursue.
The political implications of this shift in the labor market and the inadequacy of existing social protections are underexplored. Existing research on the gig workforce is largely confined to description, and rarely connected to theories about insecure labor status and political opinions and behaviors (see Traub Reference Traub2021; Zipperer et al. Reference Zipperer, McNicholas, Poydock, Schneider and Harknett2022). As such, there are still major gaps in our understanding of whether and how gig workers’ experiences shape their political opinions and participation. Perhaps relatedly, political parties and campaigns have yet to treat gig workers (or the “self-employed” more broadly) as an important constituency. In a recent column, liberal commentator Will Norris (Reference Norris2025) laments:
Yet, confoundingly, Democrats almost never talk about how their agenda has helped the self-employed. In fact, they almost never talk about independent workers as a discrete group at all. … Democratic campaign professionals slice and dice the electorate in all sorts of ways to better understand important demographics, but seldom do their polls contain cross-tabs for “self-employed.” “It just hasn’t occurred to them,” says the liberal political analyst Michael Podhorzer.
Informed by the literature on labor market dualization and precarious work (Chueri and Busemeyer Reference Chueri and Busemeyer2025; Davidsson and Emmenegger Reference Davidsson, Emmenegger, Bonoli and Natali2012; Emmenegger et al. Reference Emmenegger, Häusermann, Palier and Seeleib-Kaiser2012; Kalleberg and Vallas Reference Kalleberg, Vallas, Kalleberg and Vallas2018; Marx Reference Marx2014a; Rueda Reference Rueda2006; Schor Reference Schor2020; Schwander and Häusermann Reference Schwander and Häusermann2013; Vlandas Reference Vlandas2013) and policy feedback (Campbell Reference Campbell2002; Jacobs and Mettler Reference Jacobs and Mettler2018; Mettler Reference Mettler2002; Pierson Reference Pierson1993; Soss and Schram Reference Soss and Schram2007), the present study seeks to fill this gap with a novel online survey of 1,490 workers, including an oversample of 810 gig workers, conducted in January of 2023. The survey captures information not only on respondents’ work status and history, but also their experiences with public policy, their beliefs about relevant policy issues, and their political participation.
Our results show clear differences between those who rely on gig work as their “main job,” on the one hand, and both traditional workers and those who see gig work only as a “side job,” on the other, the latter two groups being more economically secure. We find that the use of poverty relief programs such as Temporary Assistance for Needy Families (TANF) and the Supplemental Nutrition Assistance Program (SNAP) is more prevalent for main-job gig workers than traditional workers (22% versus 7% and 54% versus 35%, respectively), but coverage by employer-provided benefits and social insurance programs is lower (e.g., 16–35% for health insurance), which aligns with existing literature (see Traub Reference Traub2021; Zipperer et al. Reference Zipperer, McNicholas, Poydock, Schneider and Harknett2022). As expected, gig workers are more in favor of expanding social programs to include them than traditional workers (0.26 standard deviations), despite an increase in taxes. They are both more knowledgeable about and more supportive of the Biden administration’s efforts to ensure classification of gig workers as employees than traditional workers. Regarding political participation, main-job gig workers are less likely to vote than traditional workers (turning out in 0.67 fewer elections out of five), but both categories of gig workers are much more likely than traditional workers to engage in nonvoting acts of participation, such as contacting an elected official or attending a march or protest (57% of main-job gig workers performed at least one such act, compared with 42% of traditional workers). Overall, the findings are consistent with our expectation that gig workers’ experiences, especially those of main-job gig workers, diverge from those of traditional workers, and that these differences have some effect on relevant political attitudes and behavior. These findings are notable, given that class consciousness and self-interested policy attitudes are far from inevitable (e.g., Bartels Reference Bartels2005; Reference Bartels2008; Lau and Heldman Reference Lau and Heldman2009; Sears et al. Reference Sears, Lau, Tyler and Allen1980).
The findings have important implications for policy makers and advocates as well as for research on welfare politics and policy feedback. First, they underscore the limitations of the existing two-tier safety net, in which the more generous contributory programs provide minimal protections for nonstandard workers, leaving them reliant on less generous means-tested programs. This mismatch highlights the need for reform to better address ongoing labor market change and the growing precarity faced by nonstandard workers. Second, the findings show that gig workers are relatively attentive and engaged on these issues, particularly in terms of nonvoting forms of activism. Drawing on new social risk theory, which emphasizes how the concerns of politically fragmented groups can still gain traction on the policy agenda, this suggests that gig workers may represent an emerging and politically relevant constituency. Yet political parties have thus far largely failed to incorporate gig workers and the “self-employed” more broadly into their electoral strategies. As such, this group holds untapped potential as a mobilizable base for efforts to expand the safety net (see Marshall Reference Marshall2025; Norris Reference Norris2025).
Labor Market Change: The Dual Labor Market and the Gig Economy
The literature on dual labor markets provides valuable insights into the politics of the gig economy. A dual labor market is divided between “insiders” and “outsiders” (Doeringer and Piore Reference Doeringer and Piore1971). The division is shaped by labor market status—whether individuals are unemployed, on temporary contracts, or employed full time—and by the level of occupational risk (Vlandas Reference Vlandas2020). Insiders are those in secure, full-time positions with comprehensive protection such as unemployment insurance (UI), while outsiders include the unemployed, temporary workers, those in precarious jobs, and individuals on nonstandard contracts who experience lower wages, reduced job security, and fewer protections (Emmenegger et al. Reference Emmenegger, Häusermann, Palier and Seeleib-Kaiser2012; Marx Reference Marx2014a; Rueda Reference Rueda2006; Schwander and Häusermann Reference Schwander and Häusermann2013; Vlandas Reference Vlandas2013). Consequently, insiders are more shielded from uncertainty, whereas outsiders face greater labor market vulnerability (Davidsson and Emmenegger Reference Davidsson, Emmenegger, Bonoli and Natali2012).
The rise of flexible and precarious work has exacerbated labor market dualization. Liberalization, characterized by the deregulation of atypical employment, has accelerated the decrease in job protection and increased reliance on nonstandard arrangements (Hinrichs and Jessoula Reference Hinrichs and Jessoula2012). As a result, insecure employment—typified by nonstandard employment and temporary work—has grown within labor markets (Kalleberg Reference Kalleberg2009; Rodgers and Rodgers Reference Rodgers and Rodgers1989; Standing Reference Standing2011). Over the past few decades, the divide between insiders and outsiders has deepened, with polarization intensifying due to labor market deregulation and welfare retrenchment (Fleckenstein and Lee Reference Fleckenstein and Lee2019). The rise of gig work further amplifies this polarization (Busemeyer and Kemmerling Reference Busemeyer and Kemmerling2020). Following the typology proposed by Vallas and Schor (Reference Schor2020), platform labor can be broadly categorized by spatial dispersion—that is, as either global or local in nature. The former, referred to as “crowdwork,” includes tasks that are mediated entirely online and can be performed from any location (e.g., Amazon Mechanical Turk), while the latter, often termed “gig work,” involves location-based services that require physical presence, such as ride hailing or food delivery (e.g., Uber and DoorDash).
This study focuses on the gig economy as a novel form of employment that reinforces labor market dualization. Gig workers, lacking formal contracts and access to social insurance, are firmly positioned as outsiders (see Schor et al. Reference Schor, Attwood-Charles, Cansoy, Ladegaard and Wengronowitz2020; Spasova et al. Reference Spasova, Ghailani, Sabato, Coster, Fronteddu and Vanhercke2021; Vallas and Schor Reference Schor2020). They exemplify “precarious” workers, characterized by limited control over their work, low wages, inadequate protection by labor laws, and restricted access to employer-provided benefits (see Hipp, Bernhardt, and Allmendinger Reference Hipp, Bernhardt and Allmendinger2015; Kalleberg and Vallas Reference Kalleberg, Vallas, Kalleberg and Vallas2018; Rodgers and Rodgers Reference Rodgers and Rodgers1989; Standing Reference Standing2011). Additionally, the unique nature of gig work—which transfers responsibility and risk onto workers who utilize their own capital (e.g., vehicles)—makes them an especially vulnerable subset of labor market outsiders (Ravenelle Reference Ravenelle2019).
Precarious but Political? How Workers with Limited Power Can Shape Welfare Politics
How do gig workers in the US engage with politics, and to what extent might they emerge as a new constituency capable of reshaping welfare politics? According to power resource theory (Esping-Andersen Reference Esping-Andersen1990; Korpi Reference Korpi1983), the expansion of traditional welfare states was driven by class-based political mobilization, particularly through the organizational power of labor unions and leftist parties. However, the historically weak labor-union presence in the US, combined with the particularly limited unionization and collective bargaining capacity among gig workers, raises important questions about the capacity of gig workers to mobilize and influence welfare-state development through conventional political channels.
While power resource theory offers a compelling explanation for the expansion of social programs in industrial societies, it has limitations when applied to the postindustrial era, where labor markets have become increasingly fragmented and the traditional working class no longer operates as a unified force. The rise of nonstandard forms of employment has contributed to the emergence of what scholars refer to as new social risks (Bonoli Reference Bonoli2005; Esping-Andersen Reference Esping-Andersen1999; Taylor-Gooby Reference Taylor-Gooby2004). New social risk theory focuses on new groups that are politically fragmented, lack strong representation, and possess limited traditional power resources, making it difficult for them to influence policy through conventional means. As Bonoli (Reference Bonoli2005) notes, these groups tend to experience low levels of political participation, weak representation in democratic institutions, and internal divisions. Compared with the industrial working class of the mid-twentieth century, new social risk constituencies are far less equipped to exert sustained political pressure. Nevertheless, under certain conditions, issues salient to these groups may enter the political agenda—not through bottom-up mobilization, but when political actors adopt them strategically, often because they are seen as potential vote-winning issues. Although the gig economy largely exists outside conventional collective bargaining structures and is often seen as unlikely to foster mobilization, recent studies have documented instances of collective actions, often driven by shared antagonism toward platform management (Cini, Maccarrone, and Tassinari Reference Cini, Maccarrone and Tassinari2022; Tassinari and Maccarrone Reference Tassinari and Maccarrone2020).
A salient example in the US is California’s Assembly Bill 5 (AB5) of 2019, which sought to classify many gig workers as employees, thereby granting them access to labor protections such as a minimum wage, overtime pay, and UI. At the federal level, the Biden administration proposed a rule in 2022 under the 1938 Fair Labor Standards Act to tighten the criteria for independent contractor status—an effort to reclassify many gig workers as employees. This proposal effectively reversed a more lenient standard adopted by the first Trump administration in 2020, which had made it easier for companies to treat workers as independent contractors. The debate drew national attention and highlighted the contested nature of gig workers’ legal and political identity. It also underscores how gig workers, despite their weak organizational power, can nonetheless shape labor and welfare policy discourse. According to new social risk theory, the extent to which policy makers and politicians seek to champion such issues more widely depends on the potential of gig workers to form a meaningful constituency. In the following sections, we discuss the unique position of gig workers relative to the social safety net, and the ways in which this may influence their opinions and participation.
Gig Workers and the Two-Tiered Safety Net
Social insurance programs consist of government programs in which workers or their employers pay dedicated taxes to the programs during the years that the workers are employed (Barnes et al. Reference Barnes, Bauer, Edelberg, Estep, Greenstein and Macklin2021). Most social insurance programs are funded by taxpayers and administered by the government, including contribution-based social protection such as Social Security, Medicare, and UI. These systems overlap with an extensive array of tax provisions and rules that subsidize or require employer-provided benefits (e.g., pensions, health insurance, family and medical leave), thereby enhancing the security of eligible workers. Taken together, these policies comprise the larger and more generous tier of the American safety net. Among these, tax-advantaged employer-provided health insurance is the most common type of health coverage in the US (Barnes et al. Reference Barnes, Bauer, Edelberg, Estep, Greenstein and Macklin2021; Keisler-Starkey and Bunch Reference Keisler-Starkey and Bunch2021). Social Security, the largest federal program in terms of cost and enrollment (Barnes et al. Reference Barnes, Bauer, Edelberg, Estep, Greenstein and Macklin2021), serves as a vital income source for older adults to alleviate old-age poverty (Romig Reference Romig2022) and also provides disability benefits to over eight million workers and their family members (Center on Budget and Policy Priorities 2024). Similarly, UI, primarily funded by employer payroll taxes levied by states and the federal government, provides a crucial protection against labor market instability.
The other, less generous, “tier” of the US safety net, known as public assistance, consists of means-tested poverty relief programs, providing comparatively limited support to those struggling to afford basic necessities such as food or housing. The two-tier system of social programs is closely connected with a segmented dual labor market, as labor market insiders are more likely to be eligible for social insurance and employer-provided benefits, while outsiders are left to rely on means-tested programs. Most social insurance systems such as employer-sponsored healthcare or retirement benefits are designed to protect workers with full-time, standard employment contracts, while significant gaps exist in the protection of workers in nonstandard employment relationships such as self-employment, temporary work, or part-time work (e.g., Forde et al. Reference Forde, Stuart, Joyce, Oliver, Valizade, Alberti, Hardy, Trappman, Umney and Carson2017). These disparities are further exacerbated by restrictive eligibility rules, which are poorly suited to a nonstandard workforce characterized by low wages, unpredictable schedules, and short-term assignments (Pilaar Reference Pilaar2018).Footnote 1 The unprecedented but temporary expansion of UI under the Coronavirus Aid, Relief, and Economic Security Act in 2020, which extended coverage to nonstandard workers, underscores the inadequacy of the current social insurance system in protecting labor market outsiders (Bae Reference Bae2025). Ultimately, the two-tiered system protects against risks such as the loss of earnings in old age, being laid off, and medical bankruptcy for labor market insiders, while providing much more limited protections for outsiders, including gig workers.
It is important to note, however, that the experiences of gig workers with economic precarity and the safety net are by no means uniform. Schor (Reference Schor2020) documents substantial differences between those who rely on gig work as their main source of income and those who use it as a supplement to another income source (such as traditional employment), with the former experiencing greater precarity and reduced bargaining power. As such, those for whom gig work is the main source of income are likely to be more reliant on public assistance, while those for whom gig work is merely a “side hustle” are more likely to have access to social insurance and employer-provided benefits.
How Labor Market Status and Public Policy Affect Political Attitudes and Participation
The literatures on policy feedback and political economy offer reasons to expect that the divergent circumstances and experiences of labor market insiders and outsiders (including gig workers) will lead to differences in political attitudes and behavior.
According to policy feedback theory, enacted policies influence the attitudes and behavior of those they affect. In terms of behavior, social welfare programs in particular have been shown to either bolster political engagement (Campbell Reference Campbell2002; Mettler Reference Mettler2002) or dampen it (Michener Reference Michener2018; Soss Reference Soss1999). Pierson (Reference Pierson1993) argues that policy feedback can occur through two distinct channels: resource effects and interpretive effects. Policies that convey resources necessary for certain forms of political participation (money, free time, civic skills) make it more likely that citizens will participate (Verba, Schlozman, and Brady Reference Verba, Schlozman and Brady1995). By contrast, interpretive effects operate by communicating messages or judgments to beneficiaries. Policy designs may shape beneficiaries’ subjective experience by communicating a sense of their role as citizens in different ways (Schneider and Ingram Reference Schneider and Ingram1993; Reference Schneider and Ingram1997). Notably, social insurance programs convey to beneficiaries that they are valued and respected members of society (Campbell Reference Campbell2002; Mettler Reference Mettler, Meyer, Jenness and Ingram2005), leading to a greater sense of political efficacy, while experiences with public assistance programs are often stigmatizing and demoralizing, leading to reduced engagement (Soss Reference Soss1999). Thus, based on their differing levels of access to social insurance and public assistance, we would expect differences in political behavior between labor market insiders and outsiders.
The policy feedback literature has also considered the effect of enacted policies on people’s attitudes or preferences regarding public policies. Compared with studies on behavioral feedback, the findings of these studies are more mixed (Campbell Reference Campbell2012, 337–38). Morgan and Campbell (Reference Morgan and Campbell2011) and Soss and Schram (Reference Soss and Schram2007) find no meaningful change in relevant attitudes following major policy changes in cash welfare and Medicare, respectively. By contrast, Mettler, Jacobs, and Zhu (Reference Mettler, Jacobs and Zhu2023) and Jacobs and Mettler (Reference Jacobs and Mettler2018) find that the Affordable Care Act (ACA) of 2010 did generate some supportive attitudinal feedbacks, but these effects were limited by partisanship and existing beliefs, factors that were only partially overcome when the threat of repeal in 2017 and 2018 appeared to activate self-interested support among some Republicans (Mettler, Jacobs, and Zhu Reference Mettler, Jacobs and Zhu2023).
The political economy literature offers some additional support for the idea that labor market position, particularly the distinction between insiders and outsiders, shapes policy preferences (Schwander and Häusermann Reference Schwander and Häusermann2013). Specifically, research shows that labor market outsiders with precarious employment tend to support more generous unemployment benefits, whereas insiders with stable and secure employment are generally less supportive of such policies (Burgoon and Dekker Reference Burgoon and Dekker2010; Häusermann, Kurer, and Schwander Reference Häusermann, Kurer and Schwander2015; Reference Häusermann, Kurer and Schwander2016; Marx Reference Marx2014a; Reference Marx2014b; Rueda Reference Rueda2005; Svalund, Saloniemi, and Vulkan Reference Svalund, Saloniemi and Vulkan2016; Vlandas Reference Vlandas2020). Additionally, Thewissen and Rueda (Reference Thewissen and Rueda2019) find that occupational risks and labor market status are important determinants of the demand for social policy. Building on this literature, Chueri and Busemeyer (Reference Chueri and Busemeyer2025) find that platform workers express stronger preferences for social investment and compensatory labor market policies than regular workers. Shifts in labor market position or heightened exposure to the risk of layoffs have also been shown to influence social policy preferences (Fossati Reference Fossati2018; Guillaud and Marx Reference Guillaud and Marx2014; Hacker, Rehm, and Schlesinger Reference Hacker, Rehm and Schlesinger2013; Iversen and Soskice Reference Iversen and Soskice2001; Levanon Reference Levanon2018; Margalit Reference Margalit2013; Naumann, Buss, and Bähr Reference Naumann, Buss and Bähr2016), further highlighting the dynamic relationship between employment conditions and political attitudes.
Of course, arguments that one’s position in the labor market can have such political effects are predicated on the idea that self-interest is an important or dominant consideration in political attitudes and behavior, and that may not be the case. In the next section, we discuss literature that suggests our expectation that gig workers’ experiences determine their politics should be viewed with skepticism.
Skeptical Perspectives on Self-Interest and Class Consciousness in Politics
A good deal of public opinion scholarship has found that the attitudes and behavior of the public diverge from the predictions of political economy and self-interest-based explanations. At the macro level, scholars have noted that middle- and low-income voters do not seem to have acted in their own best interests in an age of growing economic inequality (Bartels Reference Bartels2008; Hicks, Jacobs, and Matthews Reference Hicks, Jacobs and Matthews2016). At the individual level, studies have found evidence for two important explanations for the apparent failure of class-based political mobilization: the inability of voters to correctly perceive their own self-interest and the importance of factors other than self-interest to policy attitudes.
Bartels (Reference Bartels2005) offers a pessimistic assessment of the public’s ability to behave in a self-interested manner in his study of public opinion on the tax cuts of the George W. Bush administration, which largely favored the wealthy. He concludes that public opinion “was ill informed, insensitive to some of the most important implications of the tax cuts, and oddly disconnected from (or misconnected to) a variety of relevant values and material interests,” and argues that “ordinary people in a state of ignorance and uncertainty orient themselves with respect to complex issues of public policy … in large part, on the basis of simple-minded and sometimes misguided considerations of self-interest” (Reference Bartels2005, 22). From this perspective, it is a mistake to assume that the average gig worker perceives their self-interested stake in social welfare or labor policies in the same way that a policy expert might.
Even if gig workers understand how policies might affect them, however, this is no guarantee that this understanding determines their policy attitudes or their vote choice. A long line of literature on political behavior and psychology has found that self-interest is frequently outweighed by other considerations, including symbolic beliefs, ideology, and sociotropic concerns (e.g., Dimick, Rueda, and Segmueller Reference Dimick, Rueda and Stegmueller2018; Legerski and Berg Reference Legerski and Berg2016; Rhodebeck Reference Rhodebeck1993; Sears et al. Reference Sears, Lau, Tyler and Allen1980). Lau and Heldman (Reference Lau and Heldman2009, 535) conclude that “in the everyday world of democratic politics, under a very wide variety of circumstances, very few citizens walk around trying to figure out what their tangible short-term material interests are.” To be sure, studies have documented notable effects of self-interest in policy attitudes, but only under specific conditions where the stakes are substantial and clear (Chong, Citrin, and Conley Reference Chong, Citrin and Conley2001; Doherty, Gerber, and Green Reference Doherty, Gerber and Green2006; Haselswerdt Reference Haselswerdt2020). The validity of our expectations about gig workers as a distinct political constituency, then, may depend on the extent to which the stakes of public policy are obvious and salient to them.
Hypotheses
Building on the theoretical background discussed above, we draw a set of testable hypotheses regarding the relationship between precarious labor market status—specifically gig work—and policy take-up, social welfare attitudes, and political participation. We begin with the expectation that gig workers who rely on such work as their primary income source will be less likely to be covered by employer-provided benefits and social insurance programs than traditional workers, but more likely to rely on means-tested programs (see Traub Reference Traub2021; Zipperer et al. Reference Zipperer, McNicholas, Poydock, Schneider and Harknett2022). This expectation does not apply to those who use gig work to supplement other income sources (Schor Reference Schor2020). This leads to our first two hypotheses:
H1: Compared with traditional workers, those who rely on gig work as their main source of income will receive less protection from employer-provided benefits and social insurance programs.
H2: Compared with traditional workers, those who rely on gig work as their main source of income will use means-tested poverty relief programs at higher rates.
Furthermore, based on the literature regarding the distinct policy preferences of labor market insiders and outsiders, we expect that gig workers will demonstrate stronger support for more inclusive benefits due to the job insecurities and vulnerabilities they face, which heighten their demand for expanded social protections. This requires gig workers to both correctly perceive their self-interest and factor this consideration into their policy attitudes—which is far from a foregone conclusion, per the skeptical literature cited above.
H3: Compared with traditional workers, gig workers will be more supportive of expanding social insurance programs to nonstandard workers.
Our theoretical framework also has implications for who or what Americans see as responsible for providing protections to different categories of workers. Here, again, we expect that gig workers perceive and consider their self-interest, which is not inevitable. Self-interested considerations suggest that gig workers should be more inclined than traditional workers to hold government or employers responsible for providing benefits for gig workers, as opposed to leaving such workers to their own devices.
H4: Compared with traditional workers, gig workers will be less likely to see gig workers as responsible for providing benefits and protections for themselves, and more likely to assign responsibility to employers or the government.
Finally, we expect that precarious work experiences and exclusion from the social insurance system will lead gig workers to have divergent levels of civic and political participation compared with traditional workers. With more resources such as free time and economic security, traditional workers will be more likely to participate in politics compared with gig workers.
H5: Compared with traditional workers, gig workers will report lower levels of political participation.
Data and Methods
While some existing survey datasets (such as those collected by the Pew Research Center, the Shift Project survey conducted by the Economic Policy Institute, and the National Employment Law Project) offer valuable insights about nonstandard workers in the US, they focus on economic issues and lack a robust set of measures of relevant political attitudes and behavior. To shed light on the political side of the gig economy, we fielded an original survey of American workers through the survey firm Forthright Access. The dataset and replication materials are available via Harvard Dataverse (Bae and Haselswerdt Reference Bae and Haselswerdt2025). The survey was in the field from January 4 to January 18, 2023, and again from February 15 to February 20, 2023.Footnote 2 The final sample of 1,552 respondents included 680 traditional W-2 service-sector employees and an oversample of 872 gig-economy workers.Footnote 3 We chose to focus on service-sector workers since the nature of their work is more comparable to that of gig workers than other categories of traditional workers. Based on Vallas and Schor’s (Reference Schor2020) typology, our survey focuses primarily on gig work. Gig workers, defined as those who currently earn money by taking jobs through platform companies such as Uber, Instacart, TaskRabbit, and so forth, were further classified based on whether they consider gig work to be their main job or side job. Among 1,490 workers (62 are missing), 302 reported that gig work was their main job (about 20.3% of the total sample and 37.3% of gig workers), while 508 described it as a side job (about 34.1% of the total sample and 62.7% of gig workers).
This survey contained questions for both types of workers on their work experience, financial security, social welfare program usage, political attitudes, and political behavior. Nonstandard workers answered additional questions on their gig-work experience. This survey was self-administered on Qualtrics, with a median completion time of 11.6 minutes. The full questionnaire is included in online appendix A. The project was initially declared exempt from full review by the University of Missouri Institutional Review Board on February 15, 2022, and the updated final version was declared exempt on December 14, 2022.
In the main text, we report and display the results of difference-of-means tests across the three worker categories. In online appendix B, we also display regressions with indicators of the worker status categories as independent variables and a battery of control variables measuring partisanship, income, education, age, gender, race, and ethnicity; these results are largely similar to the uncontrolled difference-of-means results. To make our findings as generalizable as possible, we constructed sampling weights based on the distribution of age, gender, race, Hispanic ethnicity, and current gig-worker status in the Pew Research Center’s American Trends Panel Wave 94 survey, which was conducted between August 23 and August 29, 2021. Following Potter and Zheng (Reference Potter and Zheng2015), we trimmed these weights to keep all values within four interquartile ranges of the median weight. The results of analyses without weights (displayed in online appendix B) are similar to the weighted results.
Results
Table 1 displays key descriptive statistics on our sample, divided by labor market status (main-job gig worker, side-job gig worker, and traditional worker). Compared with traditional workers, both categories of gig workers are more aligned with the Democratic Party, younger, and more racially and ethnically diverse. Interestingly, though, side-job gig workers report higher levels of education and income than either main-job gig workers or traditional workers, who are indistinguishable from each other on those dimensions. This similarity is likely owing to the fact that service-sector jobs are relatively low paying, even for traditional W-2 workers (O’Herron and Schneider Reference O’Herron and Schneider2023). There is also a stark difference in the gender breakdown, with over half of side-job gig workers identifying as male compared with less than 44% for both main-job gig workers and traditional workers. Taking on gig work as a side job is more feasible for men, who generally have more free time outside their primary job because they spend less time on housework and caregiving than women (Aragão Reference Aragão2023). This distinction also aligns with Schor’s (Reference Schor2020) observations regarding those who use gig work as a primary versus supplemental income source.
Table 1 Descriptive Statistics
Note: Each cell contains the mean with the standard deviation in parentheses. The excluded category for race is “others,” which includes the following groups: Native Hawaiian or Pacific Islander, Native American, other race, and answer withheld.
Use of Social Programs
Before exploring the political beliefs and behavior of these categories of workers, we begin by considering the differences in their circumstances, especially pertaining to financial security and the safety net. Figures 1 and 2 display levels of reported uptake or coverage of social policies by the three categories of workers in the form of subgroup proportions with 95% confidence intervals. The first explores coverage by employer-provided benefits that are mandated or subsidized by public policy (employer-sponsored health insurance, worker’s compensation, retirement plans, unpaid leave under the 1993 Family and Medical Leave Act, and paid family leave), while the second covers means-tested public programs for the poor (TANF, SNAP, and Medicaid), as well as health insurance purchased through a state exchange under the ACA. While the exchanges are not targeted toward the extremely poor like the other policies, they are intended to provide affordable subsidized coverage for people above the poverty line without access to employer-provided benefits, which makes them a potentially important source of aid for nonstandard workers.Footnote 4
Figure 1 Coverage of Social Policies by Categories of Workers
Note: Sampling weights used. * p < 0.1; ** p < 0.05; *** p < 0.01.
Figure 2 Coverage of Means-Tested and ACA Programs by Categories of Workers
Note: Sampling weights used. * p < 0.1; ** p < 0.05; *** p < 0.01.
For main-job gig workers relative to traditional workers, the results are exactly as predicted by H1 and H2: they report substantially lower rates of coverage by employer-provided benefitsFootnote 5 (with most differences being statistically significant), and significantly higher usage of means-tested poverty relief programs, though not necessarily ACA-exchange insurance.Footnote 6 The picture for side-job gig workers is less straightforward. In some cases (e.g., TANF and SNAP usage) they fall between main-job gig and traditional workers, as one might expect, but in terms of employer-provided benefits, they appear as well off as traditional workers, if not better off. This is consistent with the descriptive results in table 1 showing that this group reports higher levels of income and education than main-job gig workers and traditional workers, and that it is more predominantly male. This pattern supports Schor’s (Reference Schor2020) argument that platform workers’ experiences differ substantially depending on their reliance on platform work as a primary source of income. However, as Joyce and colleagues (Reference Joyce, Stuart, Forde, Valizade, Lewin and Gollan2019) emphasize, access to social protection cannot be solely determined by whether platform workers have another job, but also by the nature of that additional employment (e.g., full-time or part-time secondary jobs). Regardless of the reasons behind the differences between main-job and side-job gig workers on these measures, these results suggest that future research would do well to distinguish between categories of gig workers rather than treating them as a monolith.
Access to social programs also appears to shape perceptions of how helpful government programs would be in overcoming crises, as shown in figure B1 in the online appendix. Main-job gig workers view government programs as significantly more helpful in overcoming a crisis than do side-job gig workers, who in turn view them as significantly more helpful than traditional workers. This pattern is somewhat surprising given findings from figures 1 and 2 that main-job gig workers are largely relegated to the less generous public assistance tier of the safety net. This may be a case of the “submerged state” (Mettler Reference Mettler2011) at work: traditional workers and side-job gig workers may not perceive the value of government support they receive, particularly those delivered through employers, whereas main-job gig workers—relying on more visible, means-tested programs—may be more likely to associate their assistance with the government.
Policy Attitudes
Do these differences in perceptions and experience drive political beliefs and attitudes about the welfare state and gig work? Before exploring policy issues specific to worker categories, we begin with a comparison of generic welfare-state attitudes as measured by a battery of five “agree/disagree” items with five-point Likert scales. The statements are “Social benefits and services make people lazy,” “Social benefits should be expanded even if it means increasing taxes,” “Government should provide generous public health insurance,” “It is an individual’s responsibility if they are unemployed,” and “The government should take action to reduce income inequality.”Footnote 7 These items scale well (Cronbach’s alpha = 0.87), so we reversed the scales for some items as appropriate and calculated the mean to create an index of prosocial welfare opinion. The mean score was 3.40 for main-job gig workers, 3.49 for side-job gig workers, and 3.34 for traditional workers, though we cannot reject the null hypothesis that mean scores are the same for main-job gig workers and the other two groups (p = 0.27 and 0.16). The only statistically significant distinction is between side-job gig workers and traditional workers (p = 0.004). Combining the two gig-work categories results in a mean welfare-opinion score of 3.45, which is significantly more liberal than the traditional worker group (p = 0.008). While more statistical power would give us more definitive results across the three subgroups, it does seem reasonable to conclude that gig workers as a whole are more in favor of a generous welfare state than traditional workers, as H3 predicts, though the difference is not large (about 0.17 sample standard deviations).
We now turn to more specific social policy questions related to different categories of workers. Figure 3 summarizes the responses of each category of respondents to a series of items on who or what should bear the primary responsibility for providing three important protections for both regular workers and nonstandard workers (such as gig workers): health insurance, UI, and workers’ compensation.Footnote 8 For each item, respondents could choose to assign primary responsibility to government, employers, nonprofit organizations or charities (abbreviated to “charity” in the graphs), and individuals themselves or their family members (abbreviated to “individuals”). Selecting one of the first two options indicates that the respondent is sympathetic to government intervention on workers’ behalf in the form of direct support or employer mandates, while selecting one of the latter two indicates that they oppose government involvement.
Figure 3 Primary Responsibility for Providing Three Social Programs
Note: Sampling weights used.
A few patterns merit discussion here. First, when asked about protections for regular workers, respondents in all categories showed an overwhelming preference for either government or employer responsibility. While main-job gig workers were perhaps a touch less sympathetic toward their traditionally employed counterparts, the “individual” and “charity” options were chosen by only small minorities across categories. Turning to nonstandard workers, the government and employer options remain relatively popular, but the consensus is less overwhelming and seems to be influenced by the respondents’ own work experiences. Substantial proportions of traditional workers answered that individuals or families should bear primary responsibility for protecting nonstandard workers from risk, with this option edging out the “employer” option for health and UI. Gig workers were less likely to select this option, and more likely to put the responsibility on the government or employers, consistent with their self-interest and H4.
Second, we note a pattern not predicted by our hypotheses: relative to protections for regular workers, respondents were more likely to hold the government responsible for each gig-worker benefit rather than employers. While all groups placed more responsibility on employers for health insurance and workers’ compensation for regular workers, the government was the far more popular choice for nonstandard workers in both cases. Respondents chose the government over employers as primarily responsible for UI for both regular and nonstandard workers. This preference could reflect the opaque design of UI, which is funded by employers but delivered as a government benefit. Even so, respondents in all categories were less likely to assign responsibility to employers for UI for nonstandard workers than for regular workers.
Figure 4 measures preferences on social policy for nonstandard workers more directly, with agree/disagree items (scaled from one to five) for two statements: “The government should expand Unemployment Insurance to cover nonstandard workers, even if it means increasing taxes,” and “The government should provide health benefits to nonstandard workers, even if it means increasing taxes.” The figure also reports results from multivariate regressions incorporating controls, which are similar to the difference-of-means results. Again, we see a divergence between gig and traditional workers, with both categories of the former significantly more in favor of both policy changes than traditional workers, consistent with H3. The differences we observe are roughly equivalent to a quarter of a sample standard deviation, a small to moderately sized effect.
Figure 4 Attitudes on Expanding Social Policies to Cover Nonstandard Workers by Categories of Workers
Note: Sampling weights used. Reg. coef. = regression coefficient, UI = unemployment insurance, HI = health insurance. * p < 0.1; ** p < 0.05; *** p < 0.01.
While the policy proposals considered in figure 4 are somewhat abstract and not on the public agenda, we also see distinctions between gig and traditional workers on a specific policy: the Biden administration’s proposed rule that would reclassify gig workers as employees.Footnote 9 Figure 5 explores respondents’ knowledge of (three-point scale) and support for (five-point scale) the proposed rule.Footnote 10 While the knowledge results reflect the relative obscurity of the proposal (with the mean scores of all three groups falling below the middle category of two, “I know a little bit about it”), main-job gig workers reported significantly higher rates of familiarity than side-job gig workers, and both reported higher rates than traditional workers. The knowledge gap between main-job gig workers and traditional workers is large: more than half of a sample standard deviation. In terms of support, the proposal is moderately popular across the three groups, but significantly more so among both categories of gig workers than among traditional workers. These differences are more modest—about one-fifth of a sample standard deviation—and statistically insignificant in the multivariate results (see table B6 in the online appendix). Along with the more abstract policy support results above, these findings suggest that experience may have affected the priorities and opinions of gig workers relative to traditional workers, consistent with H3. Still, the somewhat lukewarm support for the rule among gig workers is notable; we return to this issue in the discussion and conclusion.
Figure 5 Biden-Rule Knowledge and Support by Categories of Workers
Note: Sampling weights used. Reg. coef. = regression coefficient. * p < 0.1; ** p < 0.05; *** p < 0.01.
Political Participation
What are the implications of these differences for political action? The left panel of figure 6 reports rates of self-reported voter turnout across five recent elections: the 2020 presidential, the 2022 midterm, and the most recent gubernatorial, state legislative, and local elections in the respondent’s geography. Consistent with H5, self-reported turnout is significantly lower among main-job gig workers compared with both side-job gig workers and traditional workers, with the difference in means of 0.42 elections equivalent to about one-fifth of a sample standard deviation (though multivariate regression leads to a different conclusion, as the coefficient for the difference between main-job gig and traditional workers is not statistically significant). This may be a result of the more stable lives that the latter two categories enjoy thanks to higher rates of social insurance protections.
Figure 6 Political Participation by Categories of Workers
Note: Sampling weights used. Reg. coef. = regression coefficient. * p < 0.1; ** p < 0.05; *** p < 0.01.
Turning to other forms of political participation, however, we see a different pattern. The right-side graph summarizes the mean number of nonvoting political participation acts that respondents report having performed in the previous three years (going back to 2020) from a list of five items: “Volunteered for an election or campaign,” “Contacted an elected official or their staff about an issue,” “Donated money to an election campaign,” “Attended a march or protest,” and “Attended a town hall meeting or other public government meeting.” The graph makes clear that these other acts are rarer than voting, but also that both categories of gig workers were significantly more likely to engage in them than traditional workers. The difference in the mean number of acts between main-job gig and traditional workers (0.32) is more than one-third of a sample standard deviation, a moderately sized difference. Examining the distribution of this participation variable shows an even starker difference: 57% of main-job gig workers and 53% of side-job gig workers engaged in at least one nonvoting act, compared with just 42% for traditional workers (p < 0.001 compared with both gig-work groups).
Considering these participation results as a whole, it appears that while main-job gig workers lag other workers in terms of the most common form of participation (voting), they may make up for this deficiency with higher rates of activism.Footnote 11 The latter could be evidence of gig-work experience driving differences in more targeted political behavior (as opposed to the more generalized political participation represented by voting). This pattern aligns with findings from the US showing that economic precarity tends to reduce turnout while increasing engagement in other forms of political participation (Ojeda, Michener, and Haselswerdt Reference Ojeda, Michener and Haselswerdt2024), and with comparative evidence from Western Europe indicating that occupational precarity raises the propensity to protest (Girardi Reference Girardi2024). This pattern is also reflected in studies showing that gig workers have engaged in political activism through various forms of mobilization (Cini, Maccarrone, and Tassinari Reference Cini, Maccarrone and Tassinari2022; Tassinari and Maccarrone Reference Tassinari and Maccarrone2020). Meanwhile, side-job gig workers are a politically active group by either definition, voting at comparable rates to traditional workers and engaging in activism at comparable rates to main-job gig workers.
Discussion and Conclusion
This study explores how increasing labor market polarization may influence the political landscape and welfare attitudes. Despite the well-documented growth of insecure employment, few studies have explored how the policy attitudes and political behavior of gig workers differ from those of traditional workers. This study provides valuable new evidence on the policy experiences of gig and traditional workers, and the way these experiences have shaped divergent attitudes and patterns of participation.
Our findings confirm that those who rely on gig work as their main job are largely unprotected by the more generous tier of American social policy (social insurance and employer-provided benefits), and use means-tested public assistance programs like SNAP at much higher rates than traditional workers (Lew, Chatterjee, and Torres Reference Lew, Chatterjee and Torres2021; National Employment Law Project 2021). Importantly, however, the same cannot be said of those who see gig work only as a side job: this group appears to be better off than main-job gig workers and even traditional workers in some cases in terms of their access to the more desirable tier of the social welfare system, perhaps because they are substantially more likely to be men, who are more likely to enjoy good benefits packages than women (Horowitz and Parker Reference Horowitz and Parker2023). These findings are consistent with Schor’s (Reference Schor2020) observation that full-time gig workers tend to face greater economic vulnerability and reduced bargaining power, as they are less able to choose their engagements selectively. When asked how helpful they believed government would be in such a situation, with our expectation being that these experiences would be reflected in respondents’ subjective reports of their precarity, respondents defied our expectations: main-job gig workers saw government as significantly more helpful than the other two groups. We speculate that this could be a result of the transparency of government responsibility for public assistance programs as compared with social insurance and employer-provided benefits (Mettler Reference Mettler2011).
In terms of policy attitudes, gig workers are more supportive of the safety net in general than traditional workers, particularly when it comes to expanding social programs to include nonstandard workers and assigning responsibility for gig-worker benefits to government or employers as opposed to individuals, families, and charity. Gig workers also report higher levels of knowledge of and support for the Biden administration’s rule on gig-worker classification than do traditional workers. On these attitudinal questions, main-job and side-job gig workers are more alike than they are different, despite the clear differences in their material circumstances. These findings are consistent with political economy research finding that workers’ positions in the labor market drive differences in relevant policy attitudes (Chueri and Busemeyer Reference Chueri and Busemeyer2025; Fossati Reference Fossati2018; Marx Reference Marx2014a; Reference Marx2014b; Schwander and Häusermann Reference Schwander and Häusermann2013), but notable given that much political behavior scholarship suggests that self-interested policy attitudes are more of an exception than a rule (e.g., Lau and Heldman Reference Lau and Heldman2009; Sears et al. Reference Sears, Lau, Tyler and Allen1980).
Of course, these attitudinal differences are not always especially pronounced. In particular, gig workers’ overall weak support for the Biden rule on worker reclassification is notable: only 25% of main-job gig workers and 19% of side-job gig workers indicated that they “strongly support” the proposal. One plausible explanation for gig workers’ reticence on the rule is concern that being classified as employees could limit their flexibility and freedom. Responses to other questions on our survey highlight how important these considerations are to gig workers: 68% indicated that liking “the freedom that this work offers” was a “major reason” for pursuing gig work, while 70% indicated the same about “the ability to control my own schedule.”Footnote 12 Regressions of Biden-rule support on these items suggest that they are negatively correlated with support, though only the schedule control variable is statistically significant in a bivariate model (see table B8 in the online appendix). While these findings are only suggestive, they relate to one of the important limitations of self-interest-based models of political behavior: consensus on exactly what is in a group’s best interests may be elusive. The findings in figure 3 on responsibility for benefits suggest that the consensus may lean toward holding government responsible (i.e., through social policy) rather than employers.
In terms of political participation, main-job gig workers vote at relatively low rates, as policy feedback theory predicts. However, both main-job and side-job gig workers engage in nonvoting forms of political participation (i.e., activism) at higher rates than traditional workers. Such a pattern would be consistent with the findings of Ojeda, Michener, and Haselswerdt (Reference Ojeda, Michener and Haselswerdt2024, 2625), who find that Americans experiencing instability (e.g., job loss) were both less likely to vote and more likely to engage in other, more targeted forms of political participation than others: “[V]oting 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.” This dynamic is also consistent with comparative evidence from Western Europe, where precarity has been found to increase protest behavior (Girardi Reference Girardi2024). Considered together with the attitudinal findings, these results suggest that these workers may form a growing constituency in favor of expansion of the social safety net and reform of labor law.
Of course, this study has important limitations. It uses cross-sectional data, meaning our findings reveal correlation but do not establish causality. While the fact that most findings reported here are robust to the inclusion of controls (including partisanship) offers some level of reassurance, it cannot rule out the possibility of endogeneity. In addition, prior research has identified important differences between online and offline platform workers in terms of gender and dependent children, educational levels, pay, and working conditions (Piasna, Zwysen, and Drahokoupil Reference Piasna, Zwysen and Drahokoupil2022). However, our study is unable to distinguish between these two groups, which limits our ability to capture potential heterogeneity in platform-work experiences. It is also important to recognize the difference between support for expanding social policies and opposition to retrenchment. In contemporary US politics, for example, some Republicans became less adamantly opposed to the ACA when policy threat and loss aversion were triggered (Mettler, Jacobs, and Zhu Reference Mettler, Jacobs and Zhu2023); however, this does not necessarily imply support for expanding the ACA. Our survey is not designed to make these distinctions, however.
This paper opens several avenues for future studies. First, panel surveys could offer insights into the dynamic relationship between gig work, policy preferences, and political participation over time. Such data would also allow researchers to assess how relevant policy changes affect gig workers’ job security and access to social programs. Additionally, although the growth of the gig economy is a global trend, the financial concerns, work experiences, and political engagement of gig workers are likely to vary across institutional contexts, pointing to the need for cross-national studies. For example, even though evidence from Europe shows that platform workers across countries face low income levels (Piasna, Zwysen, and Drahokoupil Reference Piasna, Zwysen and Drahokoupil2022, 46), in countries with more universal benefits, gig workers may face less uncertainty and anxiety about future risks (see Fersch Reference Fersch2012). As Chueri and Busemeyer (Reference Chueri and Busemeyer2025) show, welfare generosity plays an important role in shaping gig workers’ social policy preferences—particularly whether they favor social investment or compensatory labor market policies.
This study also offers strategic implications for policy makers and political parties. First, this study illustrates how new social groups such as gig workers engage with policy and politics. Although regulatory efforts to enhance job security, such as California’s AB5, have faced significant setbacks—including AB5’s repeal through Proposition 22 under strong opposition from platform companies—these cases underscore a key insight from new social risk theory: issues affecting politically weak and fragmented groups can still gain visibility on the public agenda when there is potential for constituency formation. One important insight is that it may be more politically feasible to pursue a direct government role in providing protections for nonstandard workers than to pursue it via regulatory approaches like California’s AB5 and the Biden rule that put the responsibility on employers.
Our findings also speak to the importance of considering nonstandard workers themselves as a potential constituency. Lindvall and Rueda (Reference Lindvall, Rueda, Emmenegger, Häusermann, Palier and Seeleib-Kaiser2012) found that when left parties in Sweden neglect labor market outsiders, these groups are more likely to disengage from politics or turn to more radical left alternatives. In the US context, aside from limited exceptions like the aforementioned AB5 and Biden rule, political parties and campaigns have largely failed to recognize gig workers—along with the broader category of self-employed and nonstandard workers—as an influential voting bloc, a pattern that reflects the political underrepresentation of these groups identified by scholars of new social risks (Bonoli Reference Bonoli2005; Taylor-Gooby Reference Taylor-Gooby2004). This oversight marks a missed opportunity for political parties, especially Democrats, to connect with a growing and economically vulnerable demographic (see Marshall Reference Marshall2025; Norris Reference Norris2025). This opens a promising research agenda: under what conditions can gig workers become a politically organized and recognized constituency? What role might the emerging forms of labor organization, such as Rideshare Drivers United (a union for Uber and Lyft drivers), play in mobilizing this group into formal political processes? Would gig workers perceive the proposed Protecting the Right to Organize Act as an opportunity to expand collective bargaining rights, or as a threat to their flexibility? Our study provides a foundation for exploring these questions and encourages further inquiry into the evolving political representation of labor market outsiders and the future of welfare politics in an increasingly fragmented labor market.
Supplementary material
To view supplementary material for this article, please visit http://doi.org/10.1017/S1537592725103411.
Acknowledgments
The authors would like to thank Lael Keiser, Joan Hermsen, and Claire Altman for their valuable feedback and insightful suggestions on earlier drafts of this manuscript. The authors also appreciate Alexandrea Ravenelle and Daniel Schneider for their helpful advice on the gig economy and survey design. The authors are also grateful to the anonymous reviewers for their constructive comments and thoughtful suggestions.
Data Replication
Data replication sets are available in Harvard Dataverse at: https://doi.org/10.7910/DVN/JSZSWV.
Open access
