Introduction

The Great Recession – the period of economic decline in the U.S. lasting from December 2007 to June 2009 (National Bureau of Economic Research 2010) – led to the highest unemployment rates in the U.S. in over 25 years. Notably, job loss disproportionately affected men: Between 2007 and 2009, millions of jobs were lost in domains that employ primarily men, such as construction and manufacturing, and, across all domains, 75 % of job loss victims were men (Boushey 2009). While the number of jobs lost during the recession is by itself alarming, men’s disproportionate job loss may be especially troubling because of the centrality of employment to the traditional male gender role. Employment and providing for family have often been considered more central to the male than the female gender role both within and outside the U.S. (Deaux et al. 1985; Eagly and Wood 1999; Gilmore 1990; Levant and Kopecky 1995). Additionally, manhood status is more precarious than womanhood status; the perception by others that one is “a real man” requires public demonstrations of masculine qualities (e.g., physical strength) and is not guaranteed once obtained (see Vandello and Bosson 2013 for a review). Thus, to the extent that gainful employment satisfies gender role expectations and indicates manhood status, loss of employment may raise men’s doubts about their gender status.

In the present investigation, we employed a quasi-experimental methodology to examine the psychological implications of job loss for U.S. men and women by addressing two specific questions. First, how do men and women believe they will be evaluated by others – on gender-relevant dimensions – due to a (real or imagined) loss of a job? (Hereafter, we refer to beliefs about how others evaluate us as metaperceptions; Laing et al. 1966.) Second, do men’s and women’s gendered metaperceptions align with how others actually perceive them? To answer these questions, we surveyed a large, national sample of men and women from the United States shortly following the recent economic recession. Our focus is on the U.S., but given that gender stereotypes show a great deal of cross-cultural consistency (Williams and Best 1990), and the Great Recession has impacted economies globally (Verick and Islam 2010), the present investigation may have relevance for cultures outside the U.S. as well. In what follows we summarize research (conducted in the U.S. unless otherwise noted) highlighting the relationship of manhood to employment and the possible implications of this relationship for people’s metaperceptions about their gender status.

Precarious Manhood and Employment

Our first question of interest concerns men’s and women’s metaperceptions regarding their gender status. Specifically, we are interested in the extent to which men and women who lose (or imagine losing) a job expect others to (1) view them as “less of a ‘real’ man [woman],” (2) evaluate them as lower on prescriptive gender-relevant traits (traits that U.S. men and women ideally ought to have; Rudman et al. 2012), and (3) evaluate them as higher on gender proscriptive gender-relevant traits (traits considered intolerable for one gender but tolerated though undesirable for the other gender; Rudman et al. 2012), as a function of job loss. Although anthropological research provides qualitative evidence that manhood is a socially prescribed, uncertain status cross-culturally (e.g., Gilmore 1990), only recently have social psychologists employed experimental methods to demonstrate the precarious nature of manhood. This work shows that, whereas people view womanhood status as a consequence of biological maturation (e.g., reaching puberty), manhood status requires biological maturation and achieved merits (e.g., demonstrating bravery; Vandello and Bosson 2013; Vandello et al. 2008). Given the male gender role’s requirement of social achievements, men may experience threats to their manhood status when they fall short of gender role prescriptions (i.e., expectations for how men and women should behave). Indeed, manhood status can be more easily threatened than womanhood status, yielding anxiety and eliciting compensatory masculine behaviors (Bosson et al. 2005, 2009; Vandello et al. 2008; Weaver et al. 2012).

Experimental investigations have used a number of manipulations to raise men’s concerns about their manhood status in other people’s eyes (see Vandello & Bosson, 2013 for a review). For example, braiding a mannequin’s hair (a stereotypically feminine task) produced greater self-conscious discomfort among men than did a gender neutral, rope-braiding task (Bosson et al. 2005). Similarly, using a stereotypically feminine, flower-scented hand lotion versus holding a stereotypically masculine power drill motivated men to prove their manhood by making more risky financial choices (Weaver et al. 2012). In other studies, men received false feedback about their performance on a difficult test of gendered knowledge: Men who learned that they scored “like a typical woman” on this test displayed more anxiety (Vandello et al. 2008) and more negative reactions toward an effeminate gay man (Glick et al. 2007) than those who learned that they scored “like a typical man.” Thus, while violating gender roles can render both men and women vulnerable to negative evaluations (Rudman and Glick 2001), men appear especially troubled by their own gender role violations.

Studies of non-student, adult samples in the U.S. (Deaux et al. 1985; Edwards 1992; Pleck 1981) and other cultures (Gilmore 1990) indicate that gainful employment, achievement status, and ability to provide for family constitute central features of the male (but not the female) gender role. However, family structures defined strictly by men as the primary breadwinners have declined in the U.S. since the 1960s and are being replaced by more egalitarian structures (Bernard 1981; Juhn and Potter 2006; Taylor et al. 2010). Additionally, historical trends show that U.S. women now earn more college degrees than men (Peter and Horn 2005), and women’s annual income is approaching (though still less than) men’s (U.S. Census Bureau 2012). These objective gains for women may suggest a shift toward more egalitarian norms and, consequently, men and women may no longer differ as much as they once did in their beliefs about the importance of employment to their gender status.

However, several features of employment may still be more central to men’s than women’s gender status. First, employment status may be desirable for men from a mate-selection perspective (see Buss 1989), as employment may signal financial resources and social status, qualities valued by women when seeking mates (Buss 1989; Sprecher et al. 1994). Moreover, men express stronger preferences than women do for positions of high social status (Pratto et al. 1994; Sidanius et al. 2000), which can be achieved through gainful employment. Men are also evaluated relatively negatively for displaying traits inconsistent with status-strivings such as modesty (Moss-Racusin et al. 2010) and for violating workplace status expectations by working for female supervisors in masculine domains (Brescoll et al. 2012). Finally, prescriptive traits for the male gender role (e.g., assertiveness and independence) are often associated with status and lucrative employment (Eagly and Steffen 1984; Prentice and Carranza 2002; Rudman et al. 2012), whereas female prescriptive traits tend to be unrelated to status. Thus, even though objective indices suggest that employment expectations are not unique to the male gender role, we propose that U.S. men will still anticipate more gender status loss following involuntary unemployment than women will, because involuntary unemployment signals loss of social status and represents a gender role violation for men more so than for women.

In sum, the conventional centrality of employment to the male gender role, the current centrality to manhood of employment-related status outcomes, and the relative precariousness of manhood status all lead to our first prediction: that U.S. men, compared to women, will report more negative gendered metaperceptions following job loss. Put another way, men as compared to women should expect to be viewed as “less of a real man,” as lower on prescriptive gender-relevant traits, and as higher on proscriptive gender-relevant traits, for losing a job.

Are Men’s Gendered Metaperceptions Biased?

Our second question of interest concerns people’s metaperceptions, or their beliefs about how others perceive them (see Laing et al. 1966). Specifically, we explored metaperceptions about the relationship between job loss and gendered evaluations, and we examined whether or not people’s metaperceptions aligned with others’ actual perceptions of someone who recently lost their job. Unemployed men may believe that their manhood status has taken a serious blow, even when others hold relatively benign views of their gender status. For example, people often overestimate how harshly others will perceive them following embarrassing, public errors (Epley and Dunning 2006; Gromet and Pronin 2009; Savitsky et al. 2001). People similarly exaggerate the prominence of their own errors in others’ eyes (Gilovich et al. 2000) and give these errors disproportionate focus without considering the many external events that also shape impressions (Ross and Sicoly 1979). Taken together, this suggests that people may overestimate the harshness of others’ judgments of them for experiencing a job loss.

Men in particular may be especially inclined to hold negative gendered metaperceptions regarding job loss. To our knowledge, no research has examined men’s metaperceptions in the domain of unemployment, but several studies indicate that U.S. men are often biased in their views of normative expectations of masculinity. For instance, men (but not women) overestimated peer disapproval for failing to respond aggressively to a personal affront (Vandello et al. 2009). In addition, men believed that aggression was more attractive to other-gender peers and more approved of by same-gender peers than their peers actually reported. Men also overestimate the degree of muscularity that is attractive to women (Olivardia et al. 2004; Pope et al. 2000). While these studies do not address metaperceptions related to employment or job loss, they suggest that men may be biased in their understanding of normative expectations in masculine-related domains.

From our perspective, the precariousness of men’s gender status might innerve in men a hyper-vigilant attentiveness to the possibility of violating gender role norms in others’ eyes. If so, men might adaptively overestimate both the level of gender role adherence that others expect, and the social backlash others will offer should men fall short of these expectations. Reflecting this logic, we expected that men’s metaperceptions about gender status loss, regardless of whether or not these metaperceptions involved a recent actual job loss or an imagined future job loss, would be exaggerated in comparison to their peers’ actual evaluations of a hypothetical male job loss victim. Specifically, we predicted that men asked to imagine how others evaluated them after a past job loss, as well as men asked to imagine how others would evaluate them after a hypothetical job loss, would expect more negative evaluations on gender-relevant dimensions than their peers would actually provide to a hypothetical male job loss victim. We did not expect women’s gendered metaperceptions following job loss to be as exaggerated as men’s (when compared to peers’ actual evaluations).

In addressing both of these research goals, we also measured and controlled for several kinds of individual differences in our analyses. Given our focus on employed and unemployed people within the United States, we expect perceptions of gender to reflect generally shared U.S. cultural beliefs and not other demographic differences in our study sample. To rule out this possibility of observed differences in beliefs about gender resulting from demographic factors, we measured and controlled for several demographic variables (see Table 1 for a summary). First, given that beliefs about gender vary across age cohorts and education levels (see Brewster and Padavic 2000), as well as races and ethnicities (e.g., Collins 2004), we controlled statistically for the influence of participant age, education level, and race/ethnicity on our outcome measures. Second, because employment status may carry more severe consequences for households with financial dependents (e.g., unemployed spouses or children), we also controlled for marital status and number of children in the household. Finally, because the currently unemployed may be more sensitive than those currently employed to the consequences of unemployment, we controlled for current employment status.

Table 1 Demographic characteristics of study sample
Full size table

Overview and Hypotheses

We surveyed a large, nationally representative sample of adults from the U.S. that included a substantial proportion of men and women who were involuntarily unemployed during the past 4 years. While the cost of acquiring a national sample placed severe restrictions on the number of questions we could ask, this limitation was offset by the increased generalizability afforded by our sample.

Our first goal involved testing whether men, as compared to women, expected others to view them more negatively on gender-relevant dimensions following a real or imagined involuntary job loss. By “gender-relevant dimensions,” we mean gender status loss (e.g., “Not a real man [woman]”) and traits that are prescriptive and proscriptive for one’s gender; we describe these more fully in the Method section. Our hypotheses are as follows:

Hypotheses 1a-1c

After controlling for participant age, education level, current employment status, race/ethnicity, marital status, and number of children in the household, men (both employed and recently unemployed) who consider their own involuntary unemployment will report more negative gendered metaperceptions on gender status loss (Hypothesis 1a), prescriptive traits (Hypothesis 1b), and proscriptive traits (Hypothesis 1c), than women(both employed and recently unemployed). Note that, in tests of these hypotheses, employed men’s and women’s metaperceptions will be based on an imagined future job loss, whereas unemployed men’s and women’s metaperceptions will be based on a recalled real job loss.

Next, to determine whether people’s gendered metaperceptions are biased, we compared recently unemployed and employed participants’ metaperceptions against people’s ratings of unemployed others. As noted previously, we theorizethat men’s precarious gender status may lead them to overestimate the level of male role norm adherence that others expect of them in general. To test this, we asked some participants to rate a hypothetical male or female victim of involuntary job loss on all of the gender-relevant dimensions, and compared these to people’s metaperceptions. We used metaperceptions from both unemployed people guessing how others saw them and employed people imagining how others would see them to examine whether biases in people’s metaperceptions differ as a function of prior experience with unemployment (though we had no theoretical reason to expect these groups to differ). We predicted the following:

Hypotheses 2a-2c

After controlling for participant age, education level, current employment status, race/ethnicity, marital status, and number of children in the household, respondents rating a hypothetical job loss victim will offer evaluations on gender status loss (Hypothesis 2a), prescriptive traits (Hypothesis 2b), and proscriptive traits (Hypothesis 2c) that are less negativerelative to employed and recently unemployed respondents’ gendered metaperceptionson these dimensions. This discrepancy, moreover, will be greater for men than women. Thus, we expected interactions of target gender and perspective (rating self, rating hypothetical person) on gender status loss (Hypothesis 2a), prescriptive traits (Hypothesis 2b), and proscriptive traits (Hypothesis 2c).

Method

Participants

Eight-hundred-sixteen respondents (see Table 1 for sample demographics) were drawn from the Knowledge Networks Web Panel (KNWP), a database of subscribers to the Knowledge Networks service in 2010. Knowledge Networks, Inc. is a custom market research company that specializes in surveying large nationally representative populations in the U.S. Households subscribing to Knowledge Networks receive internet access in exchange for completing online surveys; thus, respondents to our survey consisted of a sample of all subscribing households who were asked by Knowledge Networks to complete the survey at their convenience. Due to the disproportionate number of employed relative to unemployed KNWP subscribers, we oversampled unemployed subscribers to achieve adequate sample sizes for hypothesis testing. The population of KNWP subscribers reflects the national population on demographics including gender, age, race, and region, based on U.S. Census data (Knowledge Networks 2012). Six participants did not answer all of the study questions, leaving a final sample of 810 participants.

Measures and Procedure

Table 2 summarizes the six cells of the experimental design (three experimental conditions for both men and women) including the number of participants per condition.

Table 2 Experimental conditions and corresponding cell sizes
Full size table

Demographic Measures

Prior to completing any survey measures for this specific study, all participants registered anonymous demographic information with Knowledge Networks. This information included age, gender, education level (measured as highest degree achieved), employment status, marital status, number of children in the household below the age of 18, and race/ethnicity (see Table 1 for summary statistics).

Employment Status and Target Ratings

Participants first indicated how many months (from 0 to 48) they were involuntarily unemployed during the past 4 years. We chose this operationalization as a way of sampling those who experienced unemployment recently enough to easily recall the experience. Thus, we classified as recentlyunemployed those who indicated more than 0 months (any involuntary unemployment during the past 4 years); we considered all others employed. Recently unemployed participants received instructions to report “how you think other people saw you at the time that you lost your job.” The survey application randomly assigned employed participants to one of two conditions: half were asked to “imagine you just lost your job” and then estimate “how others would see you.” Based on random assignment, the remaining half imagined either a hypothetical man or a woman “who very recently lost his (her) job” and was thus involuntarily unemployed; these latter participants evaluated this hypothetical target and thus provided a third-person perspective in contrast to the previous two conditions which assessed first-person perspectives (see Table 2). Participants then rated the target (self or other), “at the time of job loss,” on the following measures of gender status loss, prescriptive traits, and proscriptive traits.

Gender Status Loss

To measure gender status loss, we created a composite based on the average of two items, “Not a real man(woman)” and “Less of a man (woman).” These were rated on scales of 1 (Not at all) to 7 (Extremely) and demonstrated sufficient internal reliability, α = 0.90 for male targets, α = 0.75 for female targets.

Trait Ratings

To compare male and female targets on the extent to which they embodied prescriptive and proscriptive traits for their gender, it was necessary to use traits that differed by target gender but that were equally gendered. Participants who rated a male target thus rated him on trait masculinity, whereas those who rated a female target rated her on trait femininity. To rate trait masculinity/femininity, participants indicated the target’s standing on prescriptive and proscriptive gender-relevant traits drawn from Rudman et al. (2012). We selected groups of traits from Rudman et al. whose average effect sizes (indicating how strongly the traits were associated with one gender versus the other; see below) were similar. Thus, the male and female traits composites measured equally strong, gender-specific, prescriptive and proscriptive traits, making them appropriate for cross-gender comparisons.

Prescriptive Traits

Prescriptive trait ratings, or qualities valued for one’s gender, included competitive, assertive, independent, and has leadership ability for male targets (average d = 0.88) and warm, sensitive to others, and supportive for female targets (average d = 0.91). Respondents rated the target’s level of each trait on scales of 1 (Not at all) to 7 (Extremely), and we averaged across trait ratings to form a prescriptive traits composite (α = 0.85 for masculine traits, α = 0.82 for feminine traits).

Proscriptive Traits

Proscriptive traits, or qualities forbidden for members of one’s own gender but tolerated for members of the other gender, included indecisive, uncertain, weak, and insecure for male targets (average d = 0.86) and controlling, intimidating, aggressive, and dominating for female targets (average d = 0.89). Respondents rated the target’s level of each trait on scales of 1 (Not at all) to 7 (Extremely), and we averaged across trait ratings to form a proscriptive traits composite (α = 0.84 for masculine traits, α = 0.84 for feminine traits).

Results

Descriptive Statistics

Means and standard deviations, separated by gender for all study measures (gender status loss, prescriptive trait ratings, proscriptive trait ratings), appear in Table 3. Chi-square tests suggest a difference in the distribution of men and women to categories of employment status, χ 2(6, N = 816) = 35.00, p < .01, and to marital status categories, χ 2(4, N = 816) = 20.86, p < .01. The distribution of men and women to age categories χ 2(4, N = 816) = 1, to education level categories, χ 2(3, N = 816) = 1.09, to race/ethnicity categories, χ 2(4, N = 816) = 2.06, and to children in the household categories, χ 2(4, N = 816) = 1, did not significantly differ, all ps > 0.50. Finally, men and women did not differ significantly on the number of months of unemployment, t(814) = 0.33, ns.

Table 3 Descriptive statistics for study measures by gender
Full size table

Hypothesis 1a-1c: Do Men Report More Negative Gendered Metaperceptions than Women after an Imagined or Recalled Job Loss?

We expected employed and recently unemployed men to report more negative gendered metaperceptions for job loss – on gender status loss (Hypothesis 1a), prescriptive traits (Hypothesis 1b), and proscriptive traits (Hypothesis 1c)– than employed and recently unemployed women, when controlling for the set of covariates. The cells included in this analysis can be seen in Table 2 (cells C,E, D, and F).

We submitted the indices of gender status loss, prescriptive traits, and proscriptive traits to a 2 (Participant gender: men, women) x 2 (Employment status: recently unemployed, employed) MANCOVA while controlling for age, education, race/ethnicity, marital status, and number of children in the household. Note that we first tested for moderation of our effects by each of the covariates, to ensure that it was statistically appropriate to covary them. No moderator effects reached significance (all ps > .08). Moreover, removing the demographic covariates from all reported analyses produced identical patterns of results. The MANCOVA produced a main effect of participant gender, Wilks’ Λ = 0.86, F(3, 544) = 27.86, p < .001. When decomposed into univariate ANCOVAs, the predicted main effect emerged such that men relative to women reported greater gender status loss following a real or imagined job loss (men: M = 2.51, SD = 1.73, women: M = 2.14, SD = 1.35), F(1, 547) = 10.18, p < .01, d = 0.24, lower ratings on prescriptive traits (men: M = 4.03, SD = 1.45, women: M = 4.97, SD = 0.96) F(1, 546) = 80.11, p < .001, d = 0.71, and higher ratings on proscriptive traits (men: M = 3.20, SD = 1.54, women: M = 2.86, SD = 1.27)F(1, 544) = 9.28, p < .01, d = 0.24. No other multivariate (or univariate) effects were significant (all ps > .12). In short, supporting Hypotheses 1a-1c, both employed men imagining a job loss and unemployed men recalling a job loss reported more negative gendered metaperceptions on all three indices compared to employed and unemployed women. Note that while the gender status loss ratings that men reported were not unfavorable in an absolute sense (mean ratings were well below the midpoint of the scales), men nonetheless reported more negative metaperceptions than women did on this variable.

Hypothesis 2a-2c: Are Men’s Gendered Metaperceptions Biased?

We expected men – more than women – to overestimate others’ denigration of them on gender status loss (Hypothesis 2a), prescriptive traits (Hypothesis 2b), and proscriptive traits (Hypothesis 2c) following their recent job loss. To test this, we compared people’s gendered metaperceptions (Table 2, cells C and E for male targets, cells D and F for female targets) to third-person evaluations of a hypothetical job loss victim (cells A1 and B1 for male targets, cells A2 and B2 for female targets). First, we submitted gender status loss, prescriptive traits, and proscriptive traits to a 2 (Target gender: male, female) x 3 (Perspective: first-person employed, first-person unemployed, third-person) MANCOVA, controlling for age, education, race/ethnicity, marital status, and number of children in the household; differing degrees of freedom in the results reported below are due to missing data on some of the demographic variables. As predicted, a gender-by-perspective interaction Wilks’ Λ = 0.97, F(6, 1592) = 4.04, p < .001, emerged and qualified main effects of gender Wilks’ Λ = 0.87, F(3, 796) = 39.10, p < .001, and perspective, Wilks’ Λ = 0.96, F(6, 1592) = 4.92, p < .001. We thus proceeded with univariate analyses.

The expected gender-by-perspective interaction did not reach significance for gender status loss, F(2, 798) = 2.21, p = .11, but the means shown in Fig. 1 were consistent with our predictions. The simple effect of perspective was significant among men, F(2, 798) = 4.90, p < .01, f = 0.11, but not among women, F(2, 798) < 1, ns. That is, both unemployed and employed men’s metaperceptions of gender status loss were more extreme than were third-person raters’ evaluations of a hypothetical male job loss victim (ps = .002 and .07). This analysis also yielded a main effect of gender, F(2, 798) = 5.68, p < .02, f = 0.12, and a marginally significant main effect of perspective, F(2, 798) = 2.96, p = .05, d = 0.09.

Fig. 1
figure 1

Metaperceptions of gender status loss following unemployment

Full size image

Because we had no reason to expect employed and unemployed men to report different metaperceptions of gender status loss associated with job loss – and, indeed, these groups did not differ, F(1, 272) < 1, ns, – we reran the above analysis but first collapsed across employment status among first-person raters. Thus, we submitted metaperceptions of gender status loss to a 2 (Target gender) x 2 (Perspective: first-person, third-person) ANCOVA. In this analysis, the predicted gender-by-perspective interaction reached significance, F(1, 800) = 4.43, p < .04, f = 0.08, and qualified a main effect of perspective, F(1, 800) = 5.07, p < .03, d = 0.16. In support of Hypothesis 2a, men (regardless of employment status) anticipated more gender status loss than third-person raters attributed to a hypothetical male victim of job loss, F(1, 800) = 9.27, p < .01, d = 0.22, whereas the effect of perspective was not significant among women, F < 1, ns.

Next, the gender-by-perspective interaction emerged for prescriptive traits, F(2, 799) = 3.09, p < .05, f = 0.09, and qualified main effects of both gender, F(1, 799) = 93.49, p < .001, d = 0.68) and perspective, F(1, 799) = 8.26, p < .001, d = 0.20. However, the interaction pattern contradicted our hypothesis: Here, the simple effect of perspective was significant among women, F(2, 799) = 10.73, p < .001, f = 0.16, but not among men, F(2, 799) = 1.48, p > .22. Both employed and unemployed women anticipated higher (more gender-typical) ratings on prescriptive feminine traits (Ms = 5.00 and 4.94, SDs = 1.07 and 0.87) than third-person evaluators gave a female job loss victim (M = 4.33, SD = 1.07).

Finally, the ANCOVA on proscriptive traits produced a main effect of gender, F(1, 798) = 27.25, p < .001, d = 0.37, but neither the gender-by-perspective interaction, F(2, 798) = 2.12, p = .12, nor the simple effect of perspective among men, F(2, 798) = 2.54, p = .08, reached significance. As above, we collapsed across employment status among first-person raters and submitted metaperceptions of proscriptive traits to a 2 (Target gender) x 2 (Perspective: first-person, third-person) ANCOVA. The gender-by-perspective interaction did not reach significance, F(1, 800) = 2.55, p = .11. Thus, Hypothesis 2a received stronger support than did Hypotheses 2b and 2c (a finding we return to in the next section).

Discussion

When considering a real or imagined job loss, men expected more negative evaluations on gender-relevant dimensions from others than women expected, and men expected more negative gendered evaluations than others actually gave a hypothetical male target who lost his job. Thus, it appears that U.S. men hold exaggeratedly negative metaperceptions of the effects of unemployment on their gender status. Women’s gendered metaperceptions displayed no such bias toward negativity. In fact, women expected higher prescriptive trait ratings following job loss than people gave a hypothetical unemployed woman. This may reflect the fact that employment is less central to the female than the male gender role, and women thus do not fear gender-based backlash for job loss (Rudman and Glick 2001).

Note, however, that men’s metaperceptions were biased relative to third-party evaluations only on gender status loss (e.g., “not a real man”) and not on gendered traits (e.g., “assertive”). Although men expect to be viewed as relatively “less manly” in the abstract following job loss, they do not anticipate changes in how they are viewed on the specific traits that define how “a real man” ought (and ought not) to behave. This finding supports existing theory that argues that men’s gender status fluctuates according to social achievements (e.g., Vandello et al. 2008). That is, men anticipate losing manhood status in others’ eyes – without also losing their standing on (presumably stable) masculinity-related traits – based on a temporary behavioral outcome (job loss). Importantly, this suggests that being “a real man” means something more than merely demonstrating prescriptive and avoiding proscriptive masculine traits. Researchers interested in gender threats might therefore benefit from including measures that capture this abstract sense of “gender status loss” that is separable from, and transcends,gendered trait ratings.

Pluralistic Ignorance and Challenging Biased Metaperceptions

The present results suggest that men’s fears of emasculation following job loss are exaggerated. Compared to third-party raters’ evaluations of a hypothetical male job loss victim, men overestimated the reduction in their manhood status that unemployment would bring. This type of collective misperception (Miller and Prentice 1994) is well documented across a number of domains, such as male aggression (Vandello et al. 2009), college student social and health risk-taking (Hines et al. 2002), and college drinking (Prentice and Miller 1993), but this is the first study to our knowledge to find evidence that men may overestimate the blow to their manhood statuscaused by unemployment. We suggest that this finding may be problematic for two related reasons. First, men’s sense of self and their emotional well-being may be informed by their beliefs about how others view them, regardless of the accuracy of these beliefs. Second, men may engage in risky or damaging compensatory behaviors as a result of their misperceptions. This latter suggestion is consistent with research indicating that men engaged in more financial risk-taking following public threats to their gender status (Weaver et al. 2010). Given that offering people accurate information to reduce collective errors can reduce the likelihood of harmful behavior (e.g., Prentice and Miller 1993), future research might thus profit from similarly challenging and correcting men’s gendered metaperceptions.

Limitations and Future Directions

Although the use of a large, national sample of U.S. adults is a strength of the present study, the methodology has several limitations that merit attention. First, we required our participants to engage in hypothetical exercises which may be inaccurate or biased (Wilson and Gilbert 2003). Due to the limitations of hypothetical role-playing and memory distortions, we cannot be sure that people are able to accurately estimate how they would evaluate hypothetical others. Follow-up research would thus benefit from tracking people’s real-time perceptions of unemployed individuals. Second, our use of a U.S. sample limits direct cross-cultural generalization. More broadly, the majority of the recent empirical evidence testing precarious manhood hypotheses (and informing the current predictions) comes from United States samples, which often employ college students and which do not necessarily generalize to other samples or cultures. To be sure, there is a great deal of consistency across cultures in terms of the contents of male and female gender roles (e.g., Gilmore 1990; Williams and Best 1990). Nonetheless, cultures differ widely in the extent to which women are expected to perform behaviors conventionally reserved for men (Glick et al. 2007). It would therefore be interesting to see if the effects reported here are even stronger in cultures characterized by less gender equality. Third, our decision to classify as “recently unemployed” anyone who was involuntarily unemployed during a 48-month time span limits more nuanced conclusions. For example, people who faced a short period of involuntary unemployment 1 year in the past may differ from those who faced the same amount of unemployment more recently. Future research should therefore consider more complex operationalizations of unemployment than we were able to achieve given the necessary brevity of our survey.

Moreover, without specific information surrounding respondents’ involuntary unemployment (e.g., type of job, former salary, reason for job loss, family structure), we cannot draw specific conclusions about which men might perceive the most severe threat to their gendered status. As one example, men who hold jobs in stereotypically feminine domains may perceive negative gendered evaluations as a consequence of employment (e.g., nursing; Roth and Coleman 2008). Given men’s pressure to avoid femininity (Kierski and Blazina 2009; Norton 1997), losing a stereotypically feminine job may actually relieve this threat, complicating the associations between unemployment and gendered metaperceptions. Thus, amore nuanced measure of the nature of people’s unemployment (beyond the duration of involuntary employment as used here) would elucidate the magnitude and scope of men’s gender status concerns following involuntary unemployment. We encourage researchers to explore contextual variables that contribute to men’s negative gendered metaperceptions as a consequence of job loss.

Finally, we note that the level of negative gendered metaperceptions reported by our participants was fairly low overall for both genders. This may have occurred for two reasons. First, we used a low-impact manipulation: Participants who reported their own gendered metaperceptions engaged in a hypothetical or retrospective exercise, as opposed to the more potent alternative of responding directly after an actual job loss. The relatively low vividness and emotional intensity of the hypothetical or retrospective exercise may therefore have elicited less exaggerated perceptions of gender denigration. Second, the degree to which job loss impacts gendered metaperceptions may simply be small in the current sample. We nevertheless found evidence for relative differences in men’s versus women’s metaperceptions, which suggests that employment plays a greater role in defining the male than female gender role overall.

Concluding Remarks

Our findings suggest that men’s continued elevated rates of unemployment in the U.S. may elicit a perceived (and somewhat biased) challenge to their gender status. In spite of aslowly recovering economy (see Goodman 2010), the current results still find relevance in the context of a profound cultural shift. Women’s workforce participation continues to grow (Bureau of Labor Statistics 2000), and women are increasingly earning higher degrees (National Center for Education Statistics, 2010), making them competitive for the types of jobs traditionally filled by men. Moreover, many jobs that continue to be outsourced to other nations are stereotypically masculine ones (e.g., manufacturing; Maudlin 2010). These trends together suggest that, even with economic recovery, men will continue to confront changes in the fundamental nature of work and its ties to masculinity. Despite this struggle, we end on an optimistic note: Men faced with the threat of job loss seem to expect worse gender penalties than they are likely to receive, a misperception that may be relatively simple to disabuse.