Do Workers Speak Up When Feeling Job Insecure? Examining Workers’ Response to Precarity During the COVID-19 Pandemic

Abstract

The COVID-19 pandemic inflicted unprecedented precarity upon workers, including concerns about job insecurity. We examine whether workers respond to job insecurity with voice, and assess the role of unions, managers, and employment arrangements in this relationship. Analyses of an original 2020 survey representative of Illinois and Michigan workers show that job insecurity is not significantly associated with voice. Further, while we find that union membership and confidence in organized labor are positively associated with voice, insecure workers are less likely to speak up than secure workers as confidence in organized labor increases. Last, we find that insecure nonstandard workers are less likely to use voice than their secure counterparts.

In March 2020, the pandemic inflicted unprecedented economic uncertainty on U.S. workers. Mandated shutdowns, skyrocketing COVID-19 caseloads, and a lack of coordinated political response caused millions to lose their jobs and instilled fear among others that their employment could be at risk. These challenges unfolded rapidly and acutely: by April 2020, the unemployment rate reached nearly 15% and by June, nearly 33 million workers were receiving unemployment insurance (Bureau of Labor Statistics, 2021a; USDOL ETA, 2021). Almost a year later, over 10 million workers remained out of work, and an additional six million were working part-time involuntarily (Bureau of Labor Statistics, 2021a). Those workers whose jobs remained intact—frontline and essential workers, for example—were not spared the anxiety that accompanied heightened and widespread job insecurity, defined as concerns about the threat to continued employment. For these workers, worries about catching the virus and the potential consequences for their employment status became inescapable realities of work (Hertel-Fernandez et al., 2020). Combined, these concerns contributed to pandemic precarity, or “material deprivation and economic anxiety resulting from the COVID-19 pandemic” (Perry et al., 2021, p.1).

At the same time, the media was replete with accounts of workers responding to pandemic precarity with voice—defined as meaningful input into decisions (Budd, 2004)—to make their work safer and more secure. Workers have spoken up directly to managers, supervisors, and colleagues about health and safety measures needed in the workplace (e.g., Andrews & Kaiser Health News, 2020). Others have engaged in collective voice, for example by writing to organizational leadership (e.g., Reyes, 2021) or organizing walkouts with colleagues (e.g., Bellafante, 2020). Essential and gig workers, the latter of whom lack protections under labor law and at workplaces, collectively held protests, sickouts, and strikes (e.g., Payday Report, 2021; Scheiber & Conger, 2020).

Still others have turned to unions to represent their concerns. Even though union density in the U.S. has decreased over the past four decades to about 10% in 2021 (Bureau of Labor Statistics, 2021b), unions have played a significant role in addressing challenges associated with COVID-19. Essential workers who belong to unions have reported better workplace practices—such as receiving personal protective equipment (PPE) and disinfecting resources—than those without representation (Hertel-Fernandez et al., 2020). Unions have won hazard pay and paid sick leave for members (e.g., Communication Workers of America, 2020), and have supported workers in nonstandard arrangements who lack access to voice via unionization, such as gig workers shopping for Instacart (Jones, 2021). At the same time, one in eight workers reported having perceived employer retaliation against them or their coworkers for raising COVID-19 related concerns (National Employment Law Project, 2020). Such outcomes underscored the importance of the extent to which workers perceive managers to be receptive to workers’ input as a determinant of voice.

This leads to the question: To what extent have workers responded to pandemic precarity with voice? We investigate this by focusing on the relationship between voice and one dimension of pandemic precarity: job insecurity. How does perceived receptiveness of managers and effectiveness of unions—which we refer to as the contextual determinants of voice—factor into these decisions? And, given that workers in different arrangements, such as gig and temporary workers, have varying access to some of these voice mechanisms, how do employment arrangements influence voice responses?

We assess these questions using an original survey representative of workers in Illinois and Michigan, the 2020 Civic Engagement and Voice Survey, conducted in July 2020. Our study contributes to our understanding of employment precarity during the pandemic in several ways. First, we show how contextual elements of voice—in particular, how workers perceive their managers and unions—influence voice decisions under conditions of job insecurity during the pandemic. Second, we analyze nonstandard work in our assessment, accounting for variation in job insecurity among workers in nonstandard employment arrangements and how it influences voice actions. Finally, our study offers a snapshot of how the macro-level consequences of the pandemic intersect with concerns over job insecurity, and response decisions workers make.

Literature

The Meaning and Consequences of job Insecurity

We define job insecurity as concerns about the threat to continued employment. Job insecurity captures an individual's perception of the threat of job loss. Notably, the subjective nature of job insecurity makes it conceptually distinct from job loss or reduced work hours, the occurrence of which brings certainty to an individual's employment status (De Witte, 2005; Lee et al., 2018; Sverke et al., 2002). Job insecurity is also a multidimensional construct, encompassing collective perceptions of insecurity in a workplace (Sora et al., 2009) as well concerns that extend beyond job loss to include declining job quality or employment standards (Låstad et al., 2015). Here, however, we focus on the individual, subjective definition of job insecurity in our analysis of pandemic precarity.

Job insecurity is rooted in a variety of sources, from a specific organizational change (e.g., organizational restructuring) to broader, macro-level social and economic changes (e.g., an economic recession). Regardless of the source, however, individuals’ perceptions of insecurity are situated in their immediate context. This means that job insecurity may fluctuate as change unfolds, underscoring its unpredictability (Breevaart et al., 2020). Likewise, its dependence on context means that more than individual-level factors shape job insecurity. Relationships with coworkers, interactions with supervisors, communication from management or unions, and work settings and arrangements are all workplace-specific factors that influence how individuals perceive job insecurity (Lee et al., 2018).

Of particular concern during the COVID-19 pandemic are the adverse consequences of job insecurity. Job insecurity adversely impacts workers’ mental and physical health, compromising their well-being, inducing stress and exhaustion, and negatively affecting their sense of personal control and autonomy (Glavin, 2013; Sverke et al., 2002). Job insecurity likewise can adversely affect workers’ attitudes towards work and their employing organization (Sverke et al., 2002) and motivate a response.

Voice as a Response to Job Insecurity

Given the adverse consequences of job insecurity, how do workers respond? What actions do they take to mitigate or cope with such uncertainty during the pandemic? Prior research establishes that individuals facing job insecurity are more likely to exit an organization (Berntson et al., 2010), or reduce job and task involvement (Cheng & Chan, 2008; Lee et al., 2018). Many of these outcomes, however, emerge from studies that evaluate insecurity in the context of a specific organizational change, such as restructuring, and not an economy-wide shock that renders exit less feasible, such as the pandemic.

In the U.S., workers’ various efforts to seek safety and security at work during the pandemic, ranging from informal, individual conversations with supervisors to well-organized collective protest, illustrate a different response: voice. Voice is defined as meaningful input into decisions (Budd, 2004). The aim of voice is to induce change—a departure from the status quo—within an organization (Dundon et al., 2004; Marchington & Wilkinson, 2000). As observed during the pandemic, voice can take on a variety of forms.

On one hand, workers who speak directly to managers and supervisors about COVID-19 concerns are exercising direct voice. This occurs when workers make demands to employers on their own, without the support of a representative body such as a union. Direct voice can occur through informal means, such as casual conversations with supervisors, or through formal practices, such as an employee survey conducted by management (Marchington & Wilkinson, 2000; Mowbray et al., 2015). Suggestions about how to reconfigure workspaces to account for social distancing as well as expressions of dissatisfaction made directly to managers or supervisors (e.g., the lack of paid sick leave) during the pandemic thus comprise direct voice.

On the other hand, workers engaging in collective action through their unions, or protests and wildcat strikes organized outside of union settings, are exercising indirect voice (Dundon et al., 2004). Likewise, channels of voice outside of workplaces—including social media, online communities, and electronic petitions (Klaas et al., 2012; Kochan et al., 2019), all widely used during the pandemic—are used for indirect voice, both collectively and individually. Indirect voice is typically conceptualized as a countervailing force in the employment relationship, addressing an imbalance of power or conflict of interests through union representation (Dundon et al., 2004). In this sense, indirect voice frequently reflects conflictual or justice-oriented voice claims (Klaas et al., 2012), as observed in essential and gig workers’ protests and strikes. As with direct voice, though, indirect voice is not limited to this orientation: representative voice mechanisms can also serve to further organizational interests, such as improving work processes (Mowbray et al., 2015).

Scholarship on voice often treats these different dimensions of the concept as distinct, isolating, for example, direct versus indirect voice, or collective versus individual voice. Yet increasingly scholars note that different forms of voice often appear as bundles and are not as rigidly isolated from one another as is sometimes assumed (Wilkinson & Fay, 2011; Wilkinson et al., 2004). Recent empirical work on voice follows a similar logic, examining these different dimensions of voice concurrently (e.g., Kochan et al., 2019). Given the varied nature of voice exercised by workers during the pandemic, we adopt this broad, inclusive view of the concept.

A different stream of scholarship supports the idea that these dimensions of voice are important as a response to job insecurity. Some research suggests that a positive relationship exists between the two, finding that job insecurity motivates voice and similar participatory behaviors (e.g., Freeman & Medoff, 1984; Sverke & Hellgren, 2001). In other words, workers view job insecurity as a challenge to overcome: employees’ proactive engagement is a strategy to preserve one's job or attempt to exert control over uncertainty (Morrison, 2011; Shoss, 2017). This idea is consistent with other research that shows that, when confronted with job insecurity, workers proactively engage with strategies such as impression management to secure their employment status (De Cuyper et al., 2014).

A significant number of studies, however, find a negative relationship between job insecurity and voice (e.g., Berntson et al., 2010; Schreurs et al., 2015) and some a nonsignificant relationship (e.g., Breevaart et al., 2010; Sverke & Goslinga, 2003). These findings build on a theory of job insecurity as a stressor, motivating responses such as withdrawal from work as a coping strategy. Concerns about the costs of exercising voice—the further depletion of one's own resources, for instance, through voice efforts, or retaliation that may result if voice claims are made to unreceptive supervisor—also account for why individuals exercise less voice when faced with job insecurity (Schreurs et al., 2015; Shoss, 2017).

In this sense, job insecurity is thus not a challenge to overcome but a stressor that workers seek to alleviate (Schreurs et al., 2015). Meta-analyses of job insecurity are generally consistent with this idea, showing that job insecurity is negatively associated with outcomes that align with the participatory nature of voice, such as job involvement (Cheng & Chan, 2008; Lee et al., 2018; Sverke et al., 2002). Thus, these mostly negative findings suggest that voice as a response to job insecurity during the pandemic may be less likely than initially expected.

• Hypothesis 1: Job insecurity will be negatively associated with voice.

Voice and its Contextual Determinants

Both job insecurity and voice are rooted in the context in which they unfold, tied to organizational climates and practices, social relationships among coworkers and managers, and shared beliefs and customs from within those groups (see, e.g., Morrison, 2011 for a discussion of voice climate). Whether individuals choose to respond to job insecurity with voice is thus situated against the backdrop of relationships with and perceptions of workplace actors like unions and managers.

The Role of Unions and Managers on Voice

Unions express collective voice or capture an indirect, representative voice mechanism (Budd, 2004; Dundon, et al., 2004). They not only articulate and promote workers’ interests to employers but also serve as an important source of support for workers, especially in periods of employment precarity (Freeman & Medoff, 1984).

Unions are also a voice mechanism independent from management; as such, workers may be less concerned about the potential costs of exercising voice via union representation, and more apt to address the kinds of threats posed to their security by the pandemic. For instance, recent analysis of Occupational Health and Safety Administration (OHSA) enforcement activity finds that unionized workers are more likely to voice concerns (i.e., file a complaint) regarding health and safety violations relative to their non-union peers. This is because unions educate their members on health and safety measures and protect them from retaliation (Sojourner & Yang, 2021). Thus, we expect that union membership will be positively associated with voice. Moreover, because decisions to exercise voice are also subjective—in this case, meaning that they are influenced by how supportive workers perceive unions to be—we also expect that workers’ assessments of unions will factor into this relationship. Therefore, we expect a positive relationship between confidence in organized labor and voice.¹

Managers, on the other hand, are significant because they are most often the target of voice claims; a recent national survey on worker voice showed that just under three-quarters of respondents directed voice claims to supervisors, while less than 20% used most other mechanisms, such as unions or grievance procedures (Kochan et al., 2019). Whether managers are perceived to be receptive and responsive to voice are key determinants of such decisions to exercise voice (Bryson et al., 2006; Morrison, 2011). Managers shape workers’ perceptions of voice utility—i.e., whether workers perceive voice will produce desired effects—based on how active managers are in responding to voice claims. Similarly, managers also shape whether voice claims are perceived to be legitimate within an organization, thus motivating voice behaviors when workers perceive their input to be valued and justified (Klaas et al., 2012).

Workers are also more apt to exercise voice when they perceive that doing so poses minimal costs (such as to their reputation) or risks (such as retaliation), communicated in part by cues from organizational leaders such as managers (Klaas et al., 2012). Fear of these risks can motivate silence, or the decision to withhold voice (Morrison, 2014). Embedded within both sets of considerations—workers’ perceptions of the usefulness and legitimacy of voice, and their safety in making their input known—are features of manager-worker relationships such as trust, which positively affect engagement in voice (Mowbray et al., 2015). Thus, workers are more apt to exercise voice when they perceive that the supervisor is receptive and responsive to input, and is unlikely to impose costs. We therefore expect a positive relationship between these potential sources of social support—union membership, confidence in organized labor, and receptive managers—and worker voice.

• Hypothesis 2a: Union membership and confidence in organized labor will each be positively associated with voice.

• Hypothesis 2b: Perceived managerial receptiveness will be positively associated with worker voice.

The Role of Unions and Managers on Voice Responses to Job Insecurity

Both unions and managers are potential sources of social support for workers facing job insecurity. Unions may do so by offering protection, reducing feelings of powerlessness in the face of uncertainty, and providing collective support in times of stress. Managers and supervisors who are receptive and responsive to workers’ concerns may provide information that alleviates concerns and reduces negative attitudes toward the employer (Sverke et al., 2006). This scholarship builds on aforementioned theories of job insecurity as a stressor, which motivates behaviors like withdrawal. Sources of social support, in other words, provide coping strategies through which the negative effects of job insecurity can be mitigated, thus buffering against adverse outcomes like reduced job satisfaction or negative attitudes.

Other work questions the extent to which buffering takes place. In their study of a large Australian employer, Dekker & Schaufeli (1995), for instance, find that social supports, captured through workers’ perceptions of confidence in managers and protection from unions, do not mitigate negative outcomes of job insecurity such as mental health complaints and withdrawal from the organization. This is because such supports are unable to reduce the concerns of insecurity directly. Workers’ underlying concerns regarding threats to their employment status remain intact even with support from these actors. In their analysis of multiple European contexts, Hellgren & Chirumbolo (2003) find similar results regarding unions; unions do not moderate the relationship between job insecurity and mental health complaints. Additionally, De Witte et al. (2008), also using European data, theorized that workers’ perceptions of union support would be negatively correlated with insecurity because of a perceived violation of the social contract between employees and unions. This perceived violation comes from the fundamental role workers expect unions to play in protecting their job security; lacking said security, workers’ support for their unions may drop.

Evidence regarding how social supports may affect the relationship between job insecurity and voice specifically is somewhat limited. Most direct tests of this relationship have been done in unionized contexts and suggest that union support does not increase the likelihood of workers responding to job insecurity with voice. Underlying explanations for this finding vary. For instance, union workers facing job insecurity are found to be less likely to raise individual direct voice (e.g., speaking up to a supervisor) than non-union workers. This is argued to be because of union workers’ increased organizational loyalty in the face of insecurity, perhaps reflecting a solidaristic sense of “being in it together” (Sverke & Hellgren, 2001). Other research suggests that exercising voice via active participation with a union—for example, by bringing concerns to a shop steward, a form of indirect voice—is an uncommon response to insecurity. This may be because unionized workers, especially those who feel insecure about their jobs, rely on the union to address concerns and do not feel the need to act themselves (Sverke & Goslinga, 2003). Taken together, these pieces of evidence and stress theories of job insecurity suggest that while union membership, confidence in organized labor, and receptive managers are positively associated with voice, they do not buffer the negative effects of job insecurity that make voice responses less likely.

• Hypothesis 3a: Even as confidence in organized labor increases, insecure workers will remain less likely to exercise voice than secure workers.

• Hypothesis 3b: Even as perception of managerial receptiveness increases, insecure workers will remain less likely to exercise voice than secure workers.

Precarity in Employment Arrangements

A notable feature of voice during the pandemic is the extent to which it has involved workers in nonstandard employment arrangements, such as strikes organized by gig workers performing work for Instacart (e.g., Scheiber & Conger, 2020). Many labored in frontline, essential roles during mandated lockdowns but were forced to contend with rising COVID-19 caseloads, fewer protections to ensure their safety, and low pay. The pandemic illustrated, in sharp relief, that nonstandard workers—who include temporary help agency workers, contract firm workers, gig workers, and independent contractors or freelancers—must grapple with both employment precarity and voice in very different organizational contexts than workers in standard arrangements.

However, existing evidence suggests that nonstandard workers are less likely to engage in voice than workers in standard arrangements (Doellgast et al., 2018). In nonstandard arrangements, relationships between workers and managers are unclear, targets of voice are multiple, and access to organizational practices and rules is uncertain. Therefore, scholars argue that such arrangements stymie voice: they render unions inaccessible due to ambiguity in employment status; introduce divisions and tensions among workers; and make organizational and managerial practices meant to foster voice less effective (Marchington et al., 2004). Because nonstandard workers find themselves in a different voice context than workers in standard arrangements, they are more likely to hold different beliefs about the utility and legitimacy of voice. Temporary and contract workers, for example, are more likely than permanent, full-time workers to report not taking action—deciding on silence—when encountering a workplace problem because of the belief that doing so would make no difference (MIT Survey on Worker Voice, 2018, cited in Riordan & Kowalski, 2021). Thus, we expect that nonstandard workers are less likely than standard workers to engage in voice during the pandemic.

We also expect that job insecurity will factor into nonstandard workers’ likelihood of engaging in voice. Existing evidence suggests that job insecurity is particularly pronounced among workers in these arrangements, despite some researchers’ contention that such workers have different expectations of security because of their self-selection into nonstandard work (e.g., De Witte, 2005). Nonstandard workers are more likely to experience unpredictability than workers in permanent arrangements (Kalleberg, 2009) and report feeling more job insecure, across different types of work and position (Lee et al., 2018). Further, evidence suggests that nonstandard workers’ job insecurity is linked to the objective nature of uncertainty inherent in their different arrangements. For instance, workers with little objective security, such as temporary help agency workers, report a higher perceived probability of job loss than nonstandard workers with relatively secure positions, such as some types of independent contractors. Both, however, evaluate their odds of losing employment to be higher than workers in permanent positions (Klandermans et al., 2010).

Further evidence regarding the nature of insecurity among nonstandard arrangements supports the assertion that they will be less likely to engage in voice in response. Although nonstandard workers perceive a greater likelihood of job loss relative to permanent workers, they also anticipate that the effects of loss will not be as severe. In part, this is argued to derive from different expectations regarding mutually understood obligations that tie permanent workers more closely to their employers, making outcomes such as decreased job satisfaction or organizational commitment much more pronounced for this group (Klandermans et al., 2010). In other words, because nonstandard workers may be less invested in their employing organizations, we expect that those who feel insecure will be even less likely to engage in voice than nonstandard workers who feel secure.

• Hypothesis 4a: Workers in nonstandard work arrangements will be less likely to engage in voice than workers in standard arrangements.

• Hypothesis 4b: Insecure workers in nonstandard arrangements will be less likely to engage in voice than secure workers in nonstandard arrangements.

Data and Methods

Data

Our study draws on original data collected as part of the 2020 Civic Engagement and Voice Survey (CEVS), a representative survey of non-institutionalized adults ages 18 to 65 in Illinois and Michigan. The survey was administered by Ipsos in July 2020, who sent out online questionnaires to the Ipsos KnowledgePanel, the largest probability-based web panel in the U.S. Ipsos recruits its panel members by using address-based sampling methods. Panel members are notified of their recruitment and assignment to a study sample by email. We first pre-tested and validated our survey instrument with approximately 200 panel members in the week prior to the main survey being distributed. To increase the response rate, the invitations for the survey were followed by two subsequent email reminders on days three and five of the 11-day field period. This led to a 57% survey completion rate. In all of our analyses, we use post-stratified weights produced by Ipsos, which incorporate geodemographic benchmarks, such as gender, age, race and ethnicity, education, and household income, from the March supplement of the U.S. Census Bureau's Current Population Survey (CPS), American Community Survey (ACS), or, in certain instances, the weighted KnowledgePanel profile data.

We collected a total of 1,285 responses, including 649 from Illinois and 636 from Michigan. Our final sample consists of 797 adults in Illinois and Michigan who have worked in any capacity for pay from March to July of 2020. We exclude 43 workers who identified themselves as self-employed business owners in their most recent job, and a small number of individuals who had missing values for our main job insecurity variable. Given our focus on capturing work-related voice throughout the pandemic, survey items repeatedly prompted respondents to think about their work situation from the beginning of the pandemic (March 2020) to when the survey was administered (July 2020), even if they had experienced a change in their work situation such as a layoff, furlough, or involuntary reduction of hours.²

Measures

Outcome variable

Worker voice

We capture COVID-19-specific voice actions with questions regarding common employer response measures aimed at addressing risks of COVID-19. We ask respondents to consider voice raised to encourage these specific employer actions prior to the time our survey was administered, as well as voice raised to encourage employer actions that were perceived as needed but had not yet occurred.

To illustrate: first, we ask respondents if their employers implemented any of the following COVID-19 response measures: remote work options, PPE, safety guidelines, paid sick leave, reassurance of job security, additional pay, and clear guidelines for communications. If yes, respondents are asked about specific voice actions taken to provide input regarding these already-implemented measures; many of these voice actions are drawn from Kochan et al.'s MIT Survey on Worker Voice (see, Kochan et al., 2019). Then, we ask whether respondents thought their employers should do a better job in implementing the same set of response measures, thus identifying a perceived need for some change. If yes, we ask once again about specific voice actions taken to encourage employer action. Taken together, these questions allow us to identify voice claims about COVID-19 responses with or without any employer action.

These voice actions include: 1 = Had a conversation with a supervisor or manager; 2 = Sought advice from coworkers or others like me; 3 = Used internal formal process at my workplace; 4 = Reported to union shop steward or tried to organize/join a union; 5 = Turned to a non-union worker organization or community organization; 6 = Joined a political protest, rally or strike; 7 = Used social media or online community, electronic petition.

We combine these two voice questions to generate our COVID-specific worker voice variable. The variable receives a value of 1 if a respondent took any of the aforementioned actions (worker voice = 1) in relation to any COVID-19 response measure, and 0 if they did not. Together, both questions are structured such that they capture voice action responding to a specific need (i.e., an objective voice need); we exclude workers who reported that they did not perceive a need for their employer to implement or do a better job implementing any response measure.

Further, consistent with the literature, we created two additional dichotomous variables to distinguish direct and indirect voice actions, assigning direct voice as 1 if respondents indicated that they had a conversation with a supervisor or manager; sought advice from coworkers; or used internal formal processes at the workplace, and indirect voice as 1 if respondents indicated that they reported to a union shop steward or tried to organize; turned to a non-union worker organization; joined a political protest, rally or strike; or used social media or online community, or an electronic petition.

Key predictors

Job insecurity

The key independent variable in our analysis is job insecurity. To capture individual perceptions of job insecurity, we ask respondents to answer the question adopted from the Quality of Working Life module of the General Social Survey: “Overall, I feel my job is secure—1 = Very true, 2 = Somewhat true, 3 = Not too true, and 4 = Not at all true.” For the purposes of our analyses, we re-coded the job insecure measure as a dummy variable that equals 1 if the respondent responded, “Not too true” and “Not at all true” and 0 if “Very true” and “Somewhat true.”

Contextual determinants of voice

In our main analyses, we investigate the extent to which contextual determinants of voice influence the relationship between job insecurity and worker voice. We first use union membership, assigning a value of 1 if respondents were members of a union or an employee association similar to a union since March 1, and 0 otherwise. We also ask about confidence in organized labor (or other worker organizations) in order to capture respondents’ perceptions of unions. We amended a Gallup Poll question that, since the 1970s, has tracked confidence in organized labor, including non-traditional worker organizations (e.g., worker associations) that have become increasingly prevalent (Kochan et al., 2019). Specifically, we ask: “In general, how much confidence do you have in organized labor or other worker organizations?—1 = Complete confidence, 2 = A great deal of confidence, 3 = Some confidence, 4 = Very little confidence, 5 = No confidence at all.” The responses were reverse-coded and had a mean of 2.77.

For the second measure of voice context, we use the managerial receptiveness construct based on a three-item scale adopted from the 2011 UK Workplace Employment Relations Survey (WERS). Respondents were asked, in relation to their employer's response to COVID-19, how good their managers are at: (1) seeking the views of employees or employee representatives; (2) responding to suggestions from employees or employee representatives; and (3) allowing employees or employee representatives to influence final decisions (1 = Excellent to 5 = Poor). The responses were reverse-coded and combined into a single continuous variable, with a mean of 3.15 and Cronbach's Alpha of 0.938.

Nonstandard work

To measure nonstandard work arrangements, we ask the following question to all respondents who have worked in any capacity for pay since March 1: “Which of the following best describes the type of employment arrangement at your current or most recent job since March 1?—1 = Employed directly by my employer, with either full-time or part-time hours; 2 = Temporary help agency worker (for instance, you are paid by a temporary help agency, whether or not the job is temporary); 3 = Contract firm worker (for instance, you are paid by a company that assigns you to work at their client company's worksite or for a single customer instead); 4 = Gig worker (you perform tasks through platforms like Uber, Instacart, Lyft, or TaskRabbit); 5 = Independent consultant or freelance worker (whether you are self-employed or receive a wage or salary; your work is typically on a task, project or client basis); 6 = Business owner (self-employed and run your own business); 7 = On-call worker (for instance, you are called into work only when needed, although you may work several days in a row). The question is adopted from the Worker Organization Study (Hertel-Fernandez et al., 2022), with clarifications intended to address misreporting of self-employment activity that plagues other survey instruments (Abraham et al., 2019) and a specific response item to separate out gig work. We construct a measure of nonstandard work by assigning 1 if the respondents reported being a temporary help agency worker, contract firm worker, gig worker, independent consultant or freelance worker, or on-call worker (nonstandard work = 1) and 0 (nonstandard work = 0) if the respondents reported being employed directly by their employer.

Controls

All models in this paper control for a set of demographic, employment, and financial characteristics of our respondents. The literature largely suggests that these factors have mixed or inconsistent effects on measures of voice comparable to those used here (Morrison, 2011).³ Findings from more recent studies (e.g., Kochan et al., 2019), however, show that the likelihood of using specific types of voice mechanisms varies by gender, race/ethnicity, and education, although overall patterns remain mixed; thus, we include such controls here. Female is coded as 1 if the respondent is female and 0 if male. Race is categorized into non-Hispanic White, non-Hispanic Black, non-Hispanic other, non-Hispanic 2 or more races, and Hispanic. Age is divided into four categories (18–32, 35–45, 45–54, and 55–65 years). Socioeconomic status is measured through three levels of educational attainment—less than high school and high school, some college, and bachelor's or advanced degree—as well as level of employment—entry level, experienced (non-manager), manager/supervisor or staff/director, and executive (SVP, VP, Department Head, President, CFO, etc). We use metropolitan area and Michigan (as opposed to Illinois) dummies and political ideology (liberal, moderate, and conservative) to account for regional and political differences across the two states during the pandemic. The analysis also includes a series of dummies for occupation (24 major group occupations closely following the BLS Standard Occupational Classification) and industry (25 sectors closely following the NAICS classification system). To account for financial characteristics, we use household income categorized into six levels (<$25,000, $25,000-$49,999, $50,000-$74,999, $75,000-$99,999, $100,000–149,999, $150,000+) and two measures that capture household financial instability. We adopt a question from the Household Pulse Survey conducted by the U.S. Census Bureau during the pandemic and ask the following questions to all respondents: “Have you, or has anyone in your household experienced a loss of employment income (e.g., due to lay-off or pay-cut)”; and “Do you expect that you or anyone in your household will experience a loss of employment income (e.g., due to lay-off or pay-cut) in the next 30 days?” For each item, we assigned a value of 1 if respondents answered yes; 0 otherwise.

A summary of demographic and financial characteristics of the sample by job insecurity is presented in Table 1. Insecure workers in our sample are slightly older, non-white, and have fewer years of formal education than secure workers. A much larger share of insecure workers is likely to report both actual and expected loss of employment income in their household (58.8% and 30.8%, respectively) than secure workers (33.4% and 9.7%, respectively).

Table 1.

Demographic and Financial Characteristics of Workers in the 2020 CEVS, by Job Insecurity.

	Overall	Secure	Insecure		Overall	Secure	Insecure
	N = 797	(N = 685)	(N = 112)		N = 797	(N = 685)	(N = 112)
Variable	(%)	(%)	(%)	Variable	(%)	(%)	(%)
Women	47.8	48.5	43.8	Household (HH) Income Insecurity
Age (years)				Loss of Employment Income in the HH	37.3	33.4	58.8
18–34	32.8	33.6	28.4	Expected loss of Employment Income in the HH	12.9	9.7	30.8
35–44	22.5	22.6	21.8	Income Quartiles
45–54	25.1	24.0	31.6	<$25,000	5.6	5.1	8.4
55–65	19.6	19.9	18.3	$25,000–$49,999	15.0	13.9	21.3
Race				$50,000–$74,999	18.0	18.4	18.8
White	71.1	72.3	64.5	$75,000–$99,999	17.1	18.0	12.3
Black	11.8	11.6	12.8	$100,000–149,999	22.8	23.1	21.5
Hispanic	10.5	5.1	2.1	$150,000+	21.0	21.6	17.7
2+ Races	2.0	8.9	19.6	Top 5 Occupations
Other	4.7	2.1	1.1	Other	14.1	13.4	18.5
Education				Office and Administrative Support	9.1	8.9	10.4
LHS/HS	27.7	24.6	45.2	Sales	8.1	8.4	6.3
Some College	32.1	32.6	29.1	Management	6.7	6.5	8.2
Bachelor's degree or higher	40.3	42.9	25.7	Education, Training, and Library	6.6	6.9	4.8
Level of Employment
Entry Level	15.6	14.3	23.2
Experienced	57.4	57.4	57.5	Top 5 Industries
Manager	24.7	25.7	19.3	Factory, Manufacturing, and Woodworking	12.6	11.5	19.2
Executive	2.2	2.6	—	Health Care	11.3	12.2	6.5
Political Ideology				Retail/Stores/Shopping	10.0	10.0	10.0
Liberal	28.8	28.5	30.2	Professional, Scientific, Technical, and Business Services	9.5	10.8	2.6
Moderate	32.2	31.8	34.8	Education and Tutoring	8.7	8.7	8.5
Conservative	39.1	39.8	35.1
Metropolitan Area	87.7	87.1	91.0
State
Michigan	42.8	42.2	46.5
Illinois	57.2	57.8	53.5

Notes. Weighted by sample weights. The sample excludes business owners and respondents who had missing values for job insecurity variable.

⁺ p < .10, *p < .05, **p < .01, ***p < .001.

Results

We begin by presenting means and standard deviations for the key variables used in our analysis. Table 2 shows that 31.1% of workers exercised any voice, whether it was to encourage employers to implement any COVID-19 response measures in the past, or to do a better job going forward, as opposed to 68.9% of workers who did not exercise voice. Most voice claims were made in the form of direct voice (96.2%) rather than indirect voice (19.1%). For each voice type, 29.9% of workers used direct voice, 5.9% used indirect voice, and 4.8% used both direct and indirect voice. In terms of job insecurity, 84.9% of workers indicated that they perceived their job to be very or somewhat secure. Managerial receptiveness measure had a mean of 3.15 overall, with secure workers reporting higher managerial receptiveness than insecure workers (3.31 and 2.24; p < .001). While a slightly higher share of insecure workers reported belonging to a union (19% for insecure workers and 17.7% for secure workers), insecure workers reported slightly less confidence in organized labor or other worker organizations. Further, about 5.2% of workers worked in nonstandard arrangements. Consistent with literature, a higher share of workers reporting job insecurity were in nonstandard work arrangements than those reporting job security (11.1% and 4.2%; p < .10).

Table 2.

Descriptive Statistics of Key Variables for Respondents Who Worked in Any Capacity for Pay, March–July 2020, Overall and by Job Insecurity.

	Overall	Secure	Insecure
	(N = 797)	(N = 685)	(N = 112)
Variable	Mean	Mean	Mean
COVID-Specific Worker Voice
Voice	31.1%	31.4%	29.3%
COVID-Specific Worker Voice, by Type
Direct	29.9%	30.3%	27.7%
Indirect	5.9%	5.7%	7.1%
Both Direct and Indirect	4.8%	4.7%	5.5%
Job Insecurity (Overall, I feel my job is secure)
1-Very true	41.7%
2-Somewhat true	43.2%
3-Not too true	9.5%
4-Not at all true	5.6%
Managerial Receptiveness (alpha = 0.9382) (1-Poor to 5-Excellent; Reverse-coded)
	3.15 (1.17)	3.31 (1.12)	2.25 (1.07)***
Union Membership (1-Union member)
	17.9	17.7	19.0
Confidence in Organized Labor or Other Worker Organizations
(1-No confidence at all to 5-Complete confidence; Reverse-coded)
	2.77 (0.84)	2.79 (0.85)	2.64 (0.80)
Nonstandard Work Arrangement
Nonstandard Worka	5.2%	4.2%	11.1%+

Notes. Weighted by sample weights. The sample excludes business owners and respondents who had missing values for job insecurity variable. Standard deviations in parenthesis. Last column denote results from t-tests showing statistical significance of the difference in means between job security and insecurity.

^a Excludes business owners for the purposes of analysis.

⁺ p < .10, *p < .05, **p < .01, ***p < .001.

Who is More Likely to Speak Up during COVID-19?

We first look at the likelihood of reporting COVID-specific voice—that is, speaking up to encourage employers to implement COVID-19 response measures—by perceptions of job insecurity. Table 2, columns (2) and (3) show the proportion of insecure versus secure workers who reported voice. While a smaller share of insecure workers (29.3%) than secure workers (31.4%) exercised voice, this difference is not statistically significant. Thus, we do not find statistically significant support for Hypothesis 1, which predicted that job insecurity will be negatively associated with voice. Likewise, we find no significant differences by job insecurity, for direct, indirect, and both direct and indirect voice. Interestingly, we see a higher share of insecure workers reporting indirect voice than secure workers (7.1% and 5.7%), though the results are insignificant and are derived from too small a sample to draw a meaningful conclusion. Table 3 confirms the descriptive results using logistic regression models and including control variables. The odds that job insecure workers exercise voice are about 0.8 times as large as the odds were for their secure counterparts (smaller by a factor of about 0.2), and the results are not statistically significant.

Table 3.

Logit Regression Predicting the Likelihood of Exercising Voice (Odds Ratios).

	(1)	(2)	(3)	(4)	(5)	(6)
Job Insecurity	0.797	0.806	0.877	0.632	0.826	0.727
	(0.263)	(0.269)	(0.290)	(0.208)	(0.270)	(0.245)
Union Membership		2.604**				2.369*
		(0.844)				(0.796)
Confidence in Organized Labor			1.301+			1.316+
			(0.177)			(0.191)
Managerial Receptiveness				0.771**		0.765**
				(0.076)		(0.076)
Nonstandard Work Arrangement					0.472	0.455
					(0.268)	(0.254)
Female	1.110	1.121	1.023	1.133	1.109	1.036
	(0.262)	(0.271)	(0.246)	(0.268)	(0.264)	(0.254)
Race (White as reference)
Black	0.927	0.858	0.860	0.850	0.925	0.720
	(0.323)	(0.303)	(0.322)	(0.307)	(0.324)	(0.283)
Hispanic	1.423	1.368	1.468	1.476	1.486	1.576
	(0.590)	(0.556)	(0.612)	(0.608)	(0.609)	(0.641)
Other	2.638+	2.391	2.770*	3.090*	2.572+	2.911+
	(1.360)	(1.325)	(1.431)	(1.613)	(1.320)	(1.635)
2+ Races	0.254+	0.229*	0.316	0.245+	0.281+	0.283
	(0.183)	(0.168)	(0.235)	(0.179)	(0.203)	(0.234)
Age (18–34 as reference)
35–44	0.857	0.879	0.911	0.906	0.860	0.994
	(0.247)	(0.258)	(0.265)	(0.270)	(0.248)	(0.307)
45–54	0.471*	0.470*	0.485*	0.476*	0.491*	0.509+
	(0.150)	(0.155)	(0.154)	(0.158)	(0.156)	(0.176)
55–65	0.764	0.791	0.790	0.800	0.794	0.897
	(0.229)	(0.242)	(0.239)	(0.247)	(0.240)	(0.286)
Education (LHS/HS as reference)
Some College	1.905+	1.895+	1.976*	1.813+	1.940*	1.919+
	(0.642)	(0.639)	(0.658)	(0.621)	(0.653)	(0.645)
Bachelor or Higher	2.256*	2.435*	2.428*	2.213*	2.345*	2.669**
	(0.853)	(0.892)	(0.902)	(0.868)	(0.896)	(0.994)
Level of Employment (Entry level)
Experienced	2.422*	2.350*	2.431*	2.415*	2.419*	2.325*
	(0.972)	(0.945)	(0.977)	(1.000)	(0.971)	(0.963)
Manager	2.568*	2.792*	2.473*	2.552*	2.522*	2.549*
	(1.152)	(1.269)	(1.121)	(1.186)	(1.139)	(1.214)
Executive	2.932	3.876+	3.126	2.887	2.906	3.866+
	(2.201)	(2.856)	(2.346)	(2.275)	(2.188)	(3.000)
HH Income Loss (1 = Yes)	1.548+	1.695*	1.503	1.587+	1.568+	1.731*
	(0.388)	(0.424)	(0.381)	(0.399)	(0.392)	(0.440)
Constant	0.000***	0.000***	0.000***	0.000***	0.000***	0.000***
	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)
N	773	771	764	772	773	762

Notes. Standard errors in parentheses. All models control for a series of demographic and financial characteristics. Due to space limitations, undisclosed in this table are the coefficients for dummies concerning metropolitan area, state, political ideology, household income, expected household income loss, occupation, and industry. All models use sample weights.

⁺ p < .10, *p < .05, **p < .01, ***p < .001.

Table 3 also presents findings regarding the relationship between contextual determinants of voice (union membership, confidence in organized labor, and managerial receptiveness) with decisions to exercise voice. Consistent with Hypothesis 2a, Model 1 shows that the odds of voice among union members are 2.6 times greater than the odds for nonunion members. In addition, the odds of exercising voice are 1.3 times greater with each additional unit increase in confidence in organized labor. Turning to managerial receptiveness, Table 3 Model (2) shows that the odds of exercising voice decreased by a factor of about 0.23 with each additional increase of perceived managerial receptiveness, which does not support Hypothesis 2b.⁴

Relationship between Job Insecurity, Voice Context, and Worker Voice

Next, we look at how these contextual determinants intersect with the relationship between job insecurity and worker voice in order to evaluate Hypotheses 3a and 3b. We theorized that, even when insecure respondents report union membership, increasing confidence in organized labor, and increasing managerial receptiveness, they will be less likely to engage in voice than secure respondents. We present the estimates of the marginal effects of job insecurity at representative values of these contextual determinants of voice in Table 4, as it is difficult to draw substantive conclusions from interaction terms in nonlinear models. We first estimate a series of logistic regression models with interaction terms, controlling for demographic and financial characteristics. Then, we show how insecurity intersects with voice context by comparing group differences in marginal effects of voice contexts and job insecurity (following, e.g., Long and Mustillo, 2021).⁵

Table 4.

Estimates of Marginal Effects of Job Insecurity from Logit Regression Models Predicting Worker Voice by Voice Context.

Voice Context	Job Secure	Job Insecure	Difference (Insecurity-Security)
A. Union Membership (N=771)
Nonunion member	0.289***	0.263***	−0.026
	(0.025)	(0.054)	(0.059)
Union member	0.485***	0.404**	−0.081
	(0.062)	(0.144)	(0.152)
B. Confidence in Organized Labor (N=764)
1-Low	0.221***	0.384**	0.163
	(0.042)	(0.131)	(0.138)
2	0.276***	0.323***	0.047
	(0.026)	(0.070)	(0.077)
3	0.339***	0.267***	−0.072
	(0.019)	(0.048)	(0.055)
4	0.407***	0.217**	−0.189*
	(0.039)	(0.076)	(0.086)
5-High	0.478***	0.174+	−0.304*
	(0.067)	(0.105)	(0.124)
C. Managerial Receptiveness (N=772)
1-Low	0.411***	0.453***	0.042
	(0.051)	(0.088)	(0.104)
2	0.373***	0.314***	−0.059
	(0.032)	(0.054)	(0.064)
3	0.337***	0.200***	−0.138*
	(0.022)	(0.051)	(0.057)
4	0.302***	0.117*	−0.185**
	(0.026)	(0.054)	(0.061)
5-High	0.269***	0.064	−0.205***
	(0.038)	(0.046)	(0.061)

Notes. Standard errors in parentheses. All models control for a series of demographic and financial characteristics. Columns (1) and (2) of this table presents predictive margins of secure and insecure workers in exercising voice at representative values of voice context. Column (3) of this table presents marginal effects at representative values (MER) of union membership, confidence in organized labor, and managerial receptiveness, with job security as the reference.

⁺ p < .10, *p < .05, **p < .01, ***p < .001.

As can be seen in Table 4, columns (1) and (2) present marginal effects across low to high levels of voice context (e.g., low to high confidence in organized labor), separately for secure workers and insecure workers. The last column shows the difference between insecure workers and secure workers at each unit of union membership status, confidence in organized labor, and perceived receptiveness of management. The results in Panel (A) show that union membership is associated with a higher probability of voice for both insecure and secure workers, and that there is no statistically significant difference between the two. Turning to confidence in organized labor, however, we find in Panel (B) column (3) that the voice-insecurity gap—i.e., the difference in probability of exercising voice between job insecure and secure workers— increases. In fact, insecure workers have a higher probability of reporting voice at lower levels of confidence in organized labor and an increasingly lower probability of reporting voice at higher levels of confidence (p < .05 at confidence in organized labor = 5). The opposite is true for secure workers: the probability of reporting voice increases with higher levels of confidence in organized labor. This shows that overall, insecure workers remain less likely to exercise voice, even as confidence in organized labor increases, providing partial support for Hypothesis 3a.

For perceived managerial receptiveness, recall that we find a negative relationship with worker voice, which is inconsistent with Hypothesis 2b. Interestingly, Panel (C) of Table 4 shows that both insecure and secure workers have a decreasing probability of exercising voice even as their perceived managerial receptiveness increases, and the probability of voice for insecure workers decreases at a much faster rate. Likewise, the difference in the probability of exercising voice between job insecure and secure workers is much larger at higher levels of perceived managerial receptiveness (column 3 of Panel C). This lends support for Hypothesis 3b, which predicted that insecure workers remain less likely to exercise voice than secure workers even as perception of managerial receptiveness increases.

Figure 1 plots the predicted probability of worker voice by job insecurity across the range of values for managerial receptiveness and confidence in organized labor. The plot shows clearly that the voice-insecurity gap increases as confidence in organized labor and perceived managerial receptiveness each increase.

Figure 1.

Predicted probability of worker voice by job insecurity and voice context.

Explaining the Lack of Voice in the Presence of Receptive Managers

Contrary to our expectation, we do not find that perceived managerial receptiveness is positively associated with worker voice. In order to understand why, we turn to a discussion of silence.

Silence is the decision to not engage in voice (Morrison, 2011, 2014). Notably, silence is not merely an absence of voice, which may simply result when individuals do not perceive the need to make a voice claim. Silence, rather, is an active decision to refrain from offering input, motivated by a sense that speaking up will not make any difference, or that managers will not value voice claims. Silence is also motivated by concerns regarding the risks, or costs, associated with voice. These can range from risks that are subjective in nature—how coworkers and supervisors might perceive those who speak up, or potential effects on reputation—to repercussions such as lost promotions or even termination. Like voice, silence is therefore contextual, dependent upon shared beliefs, social relationships, and organizational climate (Morrison, 2011, 2014).

In our survey, we asked a subset of our respondents who indicated, “I did not take any action” to encourage employers to implement COVID-19 response measures about their decision to choose silence. Specifically, we asked: “Please tell us more about why you did not take any action to encourage your employer to implement the COVID-19 response measures.” Table 5 details three key reasons for choosing silence (respondents could select as many as were applicable): (a) “My employer/management initiated the measure(s) in response to COVID-19 proactively”; (b) “I was afraid that taking an action would hurt my career or future in this job”; and (c) “I didn't think any action I could take would make a difference.”

Table 5.

Reasons for Silence, by Managerial Receptiveness and Job Insecurity.

	Overall (N = 619)	Secure (N = 534)	Insecure (N = 85)
All
(a) “My employer/management initiated the measure(s) in response to COVID-19 proactively”	60.9	65.5	34.8***
(b) “I was afraid that taking an action would hurt my career or future in this job.”	5.5	4.2	13.3*
(c) “I didn't think any action I could take would make a difference.”	18.3	16.7	27.6
Managerial Receptiveness < = 2.5 (Low)
(a) “My employer/management initiated the measure(s) in response to COVID-19 proactively”	40.5	51.4	18.0***
(b) “I was afraid that taking an action would hurt my career or future in this job.”	15.2	13.1	19.5
(c) “I didn't think any action I could take would make a difference.”	32.5	31.0	35.6
Managerial Receptiveness >2.5 (High)
(a) “My employer/management initiated the measure(s) in response to COVID-19 proactively”	69.2	69.6	64.0
(b) “I was afraid that taking an action would hurt my career or future in this job.”	1.6	1.5	2.5
(c) “I didn't think any action I could take would make a difference.”	12.6	12.5	13.7

Notes. Weighted by sample weights. Managerial receptiveness is measured in a scale of 1 to 5. Last column denote results from t-tests showing statistical significance of the difference in means between job security and insecurity.

⁺ p < .10, *p < .05, **p < 0.01, ***p < .001.

Overall, 18.3% of workers reported that they didn't think their action could make a difference, and only 5.5% of workers reported fears of potential retaliation for the reason for silence. A much higher share of workers—60.9%—reported that they chose not to engage in voice due to having employers/management who proactively initiated COVID-19 response measures.

Then, we look at whether these reasons for silence vary by perceived managerial receptiveness. We find that respondents who perceived high managerial receptiveness (>2.5) are more likely to report that their employers were proactive in implementing COVID-19 response measures relative to respondents who perceived low managerial receptiveness (≤2.5) (69.2% as opposed to 40.5%, respectively). To a lesser extent, those who perceived high managerial receptiveness reported concerns about voice utility (12.6%) or retaliation (1.6%); these reasons for silence were more common among those with low perceived managerial receptiveness (32.5% and 15.2%, respectively). These results suggest that the unexpected negative relationship we found between perceived managerial receptiveness and voice is partially due to employers who proactively pursued COVID-19 response measures.

The reason for silence also varies by perceived job insecurity. Overall, a smaller portion of insecure workers (34.8%) gave proactive employers as their reason for silence, compared to 65.5% of secure workers (p < .001). Proactive employers were even less likely to be a reason for silence among insecure workers with low levels of perceived managerial receptiveness than their secure counterparts (18.0% and 51.4%, respectively (p < .001)). However, a larger share of insecure workers than secure workers reported fears of potential retaliation (13.3% and 4.2%, respectively; p < .05). Although this difference loses significance at high levels of managerial receptiveness, it is nonetheless consistent with the underlying idea that, even with receptive managers, the stressor of job insecurity may induce silence.

The Relationship between Job Insecurity, Employment Arrangement, and Worker Voice

Finally, we investigate the relationship between job insecurity, voice, and employment arrangement. As discussed previously, our descriptive results on Table 2 show that a higher share of nonstandard workers perceived their jobs to be insecure than standard workers, which is consistent with previous evidence that job insecurity is more pronounced among workers in nonstandard employment arrangements (e.g., Klandermans et al., 2010). Turning to voice, Table 3 shows that the odds of workers in nonstandard work arrangements engaging in voice are smaller by a factor of about 0.54, consistent with Hypothesis 4a, but the results are not statistically significant. Essentially, this means that, during the COVID-19 pandemic, we observe no statistically significant difference in the likelihood of speaking up between nonstandard and standard workers.

Next, we look at whether job insecurity among nonstandard workers intersects with this relationship to voice—whether job insecurity, in other words, influences voice decisions differently for standard and nonstandard workers. We find that, among nonstandard workers, insecure workers had a much lower probability of exercising voice than secure workers, supporting Hypothesis 4b.⁶ To present the substantive effect, Figure 2 plots predicted probabilities of job insecurity on worker voice by employment arrangement. The predicted probabilities plot shows that for standard workers, there is basically no difference in the probability of exercising voice by job insecurity.

Figure 2.

Probability of worker voice by employment arrangement and job insecurity.

Discussion

In this study, we sought to understand the relationship between job insecurity and voice, given that both are notable features of the COVID-19 pandemic. We find that job insecurity is negatively but not significantly related to voice. Although we do not find support for our hypothesis, our findings are in line with some studies that found nonsignificant relationship between insecurity and voice. Because both job insecurity and voice are context-dependent, we probed the effect of workers’ beliefs about the receptiveness of their managers to voice and their confidence in organized labor, in addition to their union membership status, all of which are theorized to support voice. We also assessed how workers’ specific employment arrangements factored into the relationship between job insecurity and voice, as nonstandard arrangements are elsewhere found to be associated with both greater job insecurity and less voice than standard arrangements. We discuss each of these in turn.

We find that union members are more likely to speak up than nonunion workers, consistent with our expectations and findings from other studies of union voice during the pandemic (e.g., Sojourner & Yang, 2021). Similarly, workers are increasingly likely to speak up as their confidence in labor increases. Surprisingly, however, we find an overall negative relationship between managerial receptiveness and voice, contrary to expectations based on the literature, which posits that managers who are receptive to input can positively influence workers’ perceptions of voice utility, legitimacy, and safety, thus encouraging voice (Klaas et al., 2012). Factoring in the reasons for silence helps explain this finding: during the pandemic, many employers proactively initiated COVID-19 response measures. This was even more likely among workers who reported higher levels of perceived managerial receptiveness, thus helping explain our observed negative finding.

Further, we look at how the contextual determinants of voice—captured through measures concerning perceptions of unions and managers—intersect with the relationship between job insecurity and voice. We find that insecure workers remain less likely to exercise voice than secure workers, even as confidence in organized labor and perceived managerial receptiveness increases, consistent with our expectations derived from stress theories of job insecurity (e.g., Dekker & Schaufeli, 1995). Finally, our analysis of employment arrangements shows that nonstandard workers report higher levels of insecurity than workers in standard arrangements, but they are no more or less likely to engage in voice, contrary to our expectations about their voice behavior. Prior literature also suggests, however, that perceptions of job insecurity vary among workers in different nonstandard arrangements, leading us to assess whether this variation affects the relationship between nonstandard work and voice. We find that insecure nonstandard workers are less likely to exercise voice than nonstandard workers who are secure, consistent with our expectations.

Limitations and Contributions

One potential limitation of our study is its cross-sectional design, which constrains our ability to assess potential issues with reverse causality, i.e., that perceptions of insecurity are not the result of voice behaviors. Based on existing research, however, a cross-sectional survey is appropriate for assessing the question of voice as a response to job insecurity during the pandemic. First, the effects of job insecurity are assumed to be more or less immediate, because its context dependence means that experienced job insecurity can change from one day to the next (Berntson et al., 2010; Breevaart et al., 2020; Sverke et al., 2006). Second, longitudinal studies of job insecurity and voice did not find evidence of reverse causality, offering further evidence of voice as a response behavior to job insecurity as opposed to the other way around (Breevaart et al., 2020). Finally, like other analyses of responses to job insecurity in the context of organizational change, our study captures workers’ voice behaviors soon after a very severe economic shock, making it reasonable to assume that voice emerged as a response (Schreurs et al., 2015).

Another limitation is the extent to which our findings are generalizable to the broader sample of the U.S. population or the conditions that differ from the COVID-19 pandemic. While we find that our sample is comparable to the full U.S. population in the CPS, workers in Illinois in Michigan were more likely to be white and educated. We also did not survey our respondents prior to the onset of the pandemic, and so are not able to assess the extent to which our findings hold under different conditions. This could theoretically change the calculus of workers deciding whether or not to engage in voice; parsing out these effects and the extent to which they affect generalizability of this study and others is a topic for future research.

In conclusion, the COVID-19 pandemic has resulted in an unprecedented labor market experience for millions of workers. Our study adds to an emerging body of research unpacking the relationship between macro-level sources of job insecurity, such as the Great Recession, and various outcomes, including worker responses (e.g., Kalleberg, 2012; Lowe, 2018). Lowe persuasively argues that these broader considerations are necessary for understanding workers’ perceptions of insecurity, given that workers generally are incurring increasing risk in the labor market. We offer a critical snapshot of how the macro-level consequences of the pandemic intersect with concerns over job security, and decisions workers make regarding how to respond.

Author Biographies

Hye Jin Rho is an Assistant Professor at Michigan State University at the School of Human Resources and Labor Relations. Her research focuses on the changing nature of work and organizations and its implications for employment processes and outcomes dictating the future of work and workers. Her recent work is to appear in Management Science.

Christine Riordan is an Assistant Professor in the School of Employment Relations, UIUC. Her research centers on voice and the changing organization of work. She recently co-authored “From Bread and Roses to #MeToo: Multiplicity, distance, and the changing dynamics of conflict in IR theory,” with A.M. Kowalski in ILR Review.

Christian Lyhne Ibsen is Associate Professor at FAOS/University of Copenhagen. His research focuses on collective bargaining, vocational education and training, and the future of work and employment relations. Recently, his work has been published in journals such as ILR Review, World Politics, Socio-Economic Review, and European Sociological Review.

J. Ryan Lamare is the Reuben G. Soderstrom professor, School of Labor and Employment Relations, University of Illinois at Urbana-Champaign. His research interests include employee voice, industrial relations and politics, and workplace dispute resolution. Recent publications include articles in ILR Review, Industrial Relations, and British Journal of Industrial Relations.

Maite Tapia is Associate Professor at Michigan State University at the School of Human Resources and Labor Relations. Her research focuses on worker voice. Her recent article is on the lack of Critical Race Theory and Intersectionality within Industrial Relations (in the ILR Review co-authored with Tamara L. Lee).