Teaching College in the Time of COVID-19: Gender and Race Differences in Faculty Emotional Labor.

The COVID-19 pandemic placed new teaching demands upon faculty that may have exacerbated existing race and gender disparities in the amount of emotional labor they perform. The present study surveyed 182 full-time tenured and tenure-track faculty from three small private liberal arts colleges to examine the effect of social and professional statuses on emotional labor (i.e., managing the expression of emotions to meet job requirements) during the emergency switch to remote instruction in spring 2020. Ordinary least squares (OLS) regression revealed that white cisgender men performed less emotional labor than Black, Indigenous, and People of Color (BIPOC) cisgender men, BIPOC cisgender women, and white cisgender women and gender non-conforming (GNC) faculty. Student demands for special favors fully mediated the relationship between intersectional race and gender identity and self-directed emotional labor and partially mediated its relationship with student-directed emotional labor. We conclude that the status shield afforded white cisgender men by their race and gender protected them from student demands that would have required them to engage in as much emotional labor as faculty with other intersectional race and gender identities during the pandemic. We discuss considering differences in emotional labor when making personnel decisions. Supplementary Information: The online version contains supplementary material available at 10.1007/s11199-021-01271-0.

Much of the research studying the effects of the COVID-19 pandemic on college and university faculty has focused on gender differences in research productivity (e.g., Andersen et al., 2020; Cui et al., 2020; King & Frederickson, 2021), including documenting how the increased demands of pandemic parenting have derailed the scholarly productivity of women who are mothers (Carpenter et al., 2021; Deryugina et al., 2021; Fox & Anderson, 2020). King and Frederickson (2021) posited that the decline in women’s relative rate of scholarly productivity may also have been a result of the challenges of teaching during the emergency switch to remote instruction in the spring of 2020, especially the extraordinarily high emotional labor demands that required faculty to regulate their feelings and exert extra energy to attend to their students’ emotional states (Boncori, 2020; Hall et al., 2020). In The Managed Heart, Hochschild (1983) argued that women have been taught to modify their feelings and their emotional displays both at home (emotion work) and at work (emotional labor) to smooth interpersonal relations and foster good feelings in others. Teaching from home, faculty may have engaged in emotion work with their partners and children (Ciciolla & Luthar, 2019), but according to Hochschild’s conceptualization, emotion work only becomes emotional labor when it is required as part of a paid job, such as college professor.

As one type of teaching-intensive institution, small liberal arts colleges (SLACs) provide a valuable context within which to study the emotional labor of faculty. Compared to faculty at research-intensive universities, professors at liberal arts colleges have higher teaching and service demands (Wolf-Wendel & Ward, 2006). As Birnbaum (1988) observed, the nurturing role faculty played at such institutions pre-pandemic was time-consuming. Therefore, faculty at SLACs may be expected to have engaged in high levels of emotional labor during the pandemic. To assess whether the amounts of self- and student-directed emotional labor at the beginning of the pandemic varied depending on the social and professional statuses of individual faculty, we surveyed tenure-stream faculty at three SLACs, one a former men’s college (FMC), one that has been co-ed since its founding (CC), and one a former women’s college (FWC).

Emotional Labor in College and University Teaching

Providing the help students needed during the emergency switch to remote instruction in spring 2020 required faculty to give more support than usual, perhaps requiring more emotional labor of them. While the emotional labor involved in college teaching has often been invisible and rarely counted in performance evaluations (El-Alayli et al., 2018; Gonzales & Griffin, 2020; Hanasono et al., 2019), it constituted a substantial component of teaching even before the pandemic (Bellas, 1999). In fact, in recent years, emotional labor may have become an increasingly critical part of faculty work (Goode et al., 2020; Mahoney et al., 2011). With a few notable exceptions (Bellas, 1999; El-Alayli et al., 2018; Goode et al., 2020; Lawless, 2018; Mahoney et al., 2011; Tunguz, 2016), emotional labor in college teaching has been understudied compared to elementary and high school teaching. The current research aims to address that gap.

Defining Emotional Labor

Hochschild (1983) defined emotional labor as “the management of feeling to create a publicly observable facial and bodily display” that aligned with what the worker’s job expected of them (p. 7). Hanasono et al. (2019) underscored that emotional labor involved workers attending to others’ emotional states as well as managing their own feelings. Thus, workers, including faculty, performed two types of emotional labor: self-directed and other-directed (El-Alayli et al., 2018; Pugliesi, 1999). Self-directed emotional labor referred to the extent to which professors altered their feelings in front of their students. While faculty engaged in other-directed emotional labor to manage the feelings of colleagues and administrators, we have limited our consideration of other-directed emotional labor to managing the feelings of students only.

Identity Taxation and Status Shield Protection from Emotional Labor Demands

Women faculty, especially BIPOC women, do not have the same shield against student demands as men faculty do. Hochschild (1983) argued that, “persons in low-status categories—women, people of color, children—lack a status shield against poorer treatment of their feelings” (p. 174). Hochschild observed, for example, that women flight attendants were more likely to be the targets of passenger complaints, anger, and frustration and to have a harder time establishing their authority and enforcing the rules. According to Hochschild, status can provide workers with the means to protect themselves emotionally from the “displaced feelings of others” (p. 163). Men’s higher status has generally protected them from having to perform as much emotional labor as women in a wide range of occupations (Bellas, 1999; Cottingham et al., 2015; Erickson & Ritter, 2001; Hochschild, 1983; Kolb, 2014).

Just as Hochschild argued that women’s weaker status shield made flight attendant a different job for women than for men, so too is college teaching. Women faculty were “less protected from affronts to their authority” (Bellas, 1999, p. 100). The same was true for BIPOC faculty (Harlow, 2003; Pittman, 2010; Turner et al., 2011), particularly BIPOC women (Kelly & Fetridge, 2012; Turner, 2002). Women faculty had to establish their authority and reduce “opportunities for students to challenge it, while at the same time establishing a warm, interactive classroom environment” (Bellas, 1999, p. 101). Yet a friendlier demeanor may weaken the status shield teachers use to deflect challenges to their authority (Colomy & Granfield, 2010). Higher status workers may cultivate their status shield by adopting an aura of authority “to discourage others from challenging or even contacting them” (Goodrum & Stafford, 2003, p. 181), thus protecting not only their emotions but also their time.

Hochschild (1983) reported that passengers were more likely to ask women flight attendants for help, such as “the handling of babies, the handling of children, the coddling of the old folks” (p. 176). Similarly, women, especially BIPOC women, may face higher work demands (Zambrana et al., 2017), a phenomenon often referred to as cultural taxation. Hirshfield and Joseph (2012) expanded the concept of cultural taxation to identity taxation to include other social identities, such as gender and sexual orientation. They described identity taxation as occurring “when faculty members shoulder any labour – physical, mental, or emotional – due to their membership in a historically marginalised group within their department or university, beyond that which is expected of other faculty members” (p. 214). They found that students and colleagues expected women faculty, for example, to take on more of the work with women students. They concluded that identity taxation created inequality in faculty workloads, particularly emotional labor. Extending the concept of status shield, this study asked whether privileged social and professional statuses, including not only race and gender but also rank and tenure, shielded faculty not just from student complaints but also from the requests for additional help that faculty with marginalized social identities were more likely to face as a result of identity taxation. White men faculty may have been able to use their privileged race and gender status not only to discourage students from complaining or challenging their authority but also to deter them from asking for special favors during the emergency switch to remote learning.

Race, Gender, Tenure, and Faculty Emotional Labor

In contrast to customer service occupations (e.g., Kang, 2010) and similar to health and human services professions (e.g., Cottingham et al., 2015), faculty have some authority over their students (Sass, 2000; Tunguz, 2016). Student ratings of instruction (SRIs) can upend that power relationship, however, especially for faculty with marginalized identities or in more vulnerable positions, such as untenured faculty or women immigrant faculty, by asking students to judge professors (Mahoney et al., 2011; Tunguz, 2016). Since colleges and universities use SRIs in personnel decisions, students’ ratings could hurt faculty careers, putting greater pressure on assistant and associate professors, especially those with marginalized identities, to manage their emotions in such a way as to get good ratings (Lawless & Chen, 2017; Wingfield, 2010; Zambrana et al., 2017). Tunguz (2016) found, however, that once faculty have earned tenure, only men reaped the benefits of less pressure to engage in emotional labor. She concluded that tenured women appeared to still feel compelled to conform to traditional gender roles.

The importance of SRIs in personnel decisions may lead untenured faculty and associate professors to engage in more emotional labor than tenured faculty and full professors. The power end-of-term evaluations give students may be even greater at teaching-intensive institutions, such as SLACs, because they weigh SRIs more heavily in high stakes personnel decisions than research-intensive universities do (Tunguz, 2016).

According to Kelly and Fetridge (2012), women “experienced students differently than did White males as evidenced in their gendered and racial interactions with students” (p. 40). Previous research (Anderson, 2010; Basow, 2000; El-Alayli et al., 2018; Sprague & Massoni, 2005) demonstrated that pre-pandemic, students expected women faculty to be more approachable, available, helpful, and nurturing than men. Over fifty years ago, Bernard (1964) labelled this expectation as academic momism. Students were more likely to make requests for special favors, such as a deadline extension or a grade change, from women than from men, thereby increasing women’s workload (El-Alayli et al., 2018). Women faculty “may find that they must take on extra burdens, such as helping students cope with stress or insecurities, having to set personal boundaries with them, or providing gentler feedback to them to avoid being perceived as excessively harsh” (El-Alayli et al., 2018, p. 137).

If women do not meet students’ additional nurturing expectations, students can evaluate them more negatively than men (Sprague & Massoni, 2005). In fact, Bennett (1982) found that even when women did meet the higher nurturing expectations by devoting more time to office hours and giving more personal attention, students still gave them lower ratings. El-Alayli et al. (2018) concluded that women faculty “must live up to professional expectations in the formal aspects of teaching while simultaneously serving as academic moms” (p. 138). These higher demands for nurturance from women faculty, especially BIPOC women, resulted in their performing more emotional labor pre-pandemic than their white male colleagues (Hanasono et al., 2019; Mahoney et al., 2011; Zambrana et al., 2017), a difference that may have grown even larger during the emergency switch to remote learning in spring 2020.

The emotional labor demands on BIPOC faculty pre-pandemic were especially high (Moore et al., 2010; Turner, 2002). BIPOC women reported students asking for more emotional support, not just normal advising and mentorship (Hirshfield & Joseph, 2012). BIPOC students often turned to BIPOC faculty for emotional support as they navigated predominantly white institutions. According to Turner (2002), the scarcity of BIPOC women led to their higher workloads engaging with students outside of class. Hirshfield and Joseph (2012) concluded that BIPOC women faculty experienced “additional identity taxation, beyond that of either female faculty or faculty of colour,” (p. 216) including “stereotypes portraying them as maternal or nurturing” (p. 220). According to Roberts (2020), as a black woman, she faced a racialized version of academic momism. Students expected her to be a mammy—“unconditionally nurturing, understanding, hardworking, and mothering” (para. 7).

Furthermore, BIPOC women needed to engage in more emotional labor pre-pandemic than white men professors in response to white students, particularly white men, who gave them less respect and were more likely to challenge their authority (Harlow, 2003; Hirshfield & Joseph, 2012; Kelly & Fetridge, 2012; Pittman, 2010). In addition, research has shown that pre-pandemic, students gave BIPOC faculty lower ratings on teaching evaluations (e.g., Littleford et al., 2010; Moore et al., 2021; Wallace et al., 2019). Allsopp (2020) expressed concern that as an untenured woman of color, her student ratings would be even harsher during the pandemic. The pandemic may have exacerbated these race and gender differences in unrecognized and unrewarded emotional labor.

Student Demands and Emotional Labor

Prior to the pandemic, teaching-intensive colleges and universities, especially regional public ones, had begun calling upon faculty to engage in what Goode et al. (2020) labelled intrusive teaching to increase retention rates, especially for first-generation students (Hanasono et al., 2019). Goode et al. (2020) defined intrusive teaching as practices that involved faculty “monitoring and intervening in their students’ emotional and social issues” that might hinder their ability to be successful in the classroom (p. 49). The COVID-19 pandemic likely enlarged this component of faculty emotional labor even further at regional public universities and may have dramatically increased it at other teaching-intensive institutions and even at research-intensive universities as more institutions became concerned about student retention. Scherer (2020) found that college type affected whether working class students were able to develop relationships with faculty. She concluded that teaching-intensive colleges with smaller faculties and smaller class sizes were more likely to reduce the barriers that prevented working class students from interacting with faculty, particularly outside the classroom.

This additional work with students was gendered. O’Meara et al. (2017) found that both students and colleagues expected women to do more teaching- and student-related work. Their data revealed that in an average week, women received more additional work requests, particularly from students, than men did. During the pandemic, students may have made even more demands on faculty, particularly faculty with the weaker status shields that come with less privileged social and professional statuses. Those higher student demands may have mediated the relationship between professor gender and emotional labor, as El-Alayli et al. (2018) found.

The Present Study

Drawing on previous research (e.g., Cottingham et al., 2015; El-Alayli et al., 2018; Tunguz, 2016), we expected social status would affect how much emotional labor a professor did in spring 2020. Specifically, those with more privileged social statuses (e.g., men, whites, heterosexuals, born in the United Sates) would report performing less emotional labor after the emergency switch to remote learning in spring 2020.

H1: White cisgender men would perform less self-directed and student-directed emotional labor than white women and gender non-conforming faculty, BIPOC men, and BIPOC women.

H2: Heterosexual faculty would perform less self-directed and student-directed emotional labor than LGBTQ faculty.

H3: Faculty born in the United States would perform less self-directed and student-directed emotional labor than those born outside the US.

Expanding Tunguz (2016) research on emotional labor with tenure as a predictor, we added rank, discipline, and type of college as professional statuses predicting emotional labor. We expected that those with a more privileged professional status would report engaging in less emotional labor after the emergency switch to remote learning in spring 2020.

H4: Tenured faculty would perform less self-directed and student-directed emotional labor than untenured faculty.

H5: Full professors would perform less self-directed and student-directed emotional labor than assistant and associate professors.

Given the gendered nature of academic disciplines and the gender composition of the three SLACs’ student bodies and faculties, we expected faculty emotional labor to vary by how male-identified the college and the disciplines were. Specifically, we expected that faculty in more male-identified disciplines would report doing less emotional labor than faculty in more female-identified disciplines after the switch to remote learning.

H6: Faculty in pre-professional programs, natural sciences, and social sciences would perform less self-directed and student-directed emotional labor than faculty in the arts and humanities.

We expected that faculty at the two more male-identified colleges would report performing less emotional labor than faculty at the more female-identified college in spring 2020.

H7: Faculty at FMC and CC would perform less self-directed and student-directed emotional labor than faculty at FWC.

Finally, we expected student demands for special favors as well as teaching resources to mediate the effect of social and professional statuses on emotional labor.

H8: Student demands would mediate the association between social and professional statuses and emotional labor such that the effect of status on emotional labor would be explained by student demands.

H9: Teaching resources would mediate the relationships between social and professional statuses and emotional labor such that that the effect of status on emotional labor would be explained by teaching resources.

In sum, heeding El-Alayli et al.’s (2018) call for future research to include race as well as gender, this study aims to contribute to the research on faculty emotional labor by examining race and gender intersectionally. In addition, we analyze disciplinary differences as El-Alayli et al. suggested as well as rank and type of college. Finally, we include access to teaching resources and student demands as possible mediating variables that may explain the link between social and professional status of faculty and emotional labor during the pandemic.

Method

Sample Context

Participants were tenure-stream faculty members recruited from three highly selective, predominately white, residential, private liberal arts colleges in upstate New York: one that has been co-ed since its founding (CC), one a former women’s college (FWC), and one a former men’s college (FMC). While CC has been coeducational for well over a century, the other two only became co-educational in the early 1970s. FWC has had a greater curricular emphasis on the more female-identified arts and humanities, and FMC has had a greater emphasis on the more male-identified natural sciences and engineering. FWC has the highest proportion of women faculty (52%) and students (60%) while FMC has the lowest (43% and 47% respectively). It also has significantly more full professors than FWC and CC (see Table 1). Both FMC and CC have fraternities, football, and National Collegiate Athletic Association Division I men’s and women’s ice hockey (all other sports are Division III); FWC does not.

Table 1

Means and (Standard Deviations) by College of Employment

	Pooled	FWC	CC	FMC
	(N = 182)	(n = 80)	(n = 35)	(n = 67)	F	df	p	η2	n
Emotional Labor
Self-directed Emotional Labor	2.91 (1.06)	2.99 (1.06)	3.16 (.10)	2.69 (1.09)	2.45	(2,168)	.090	.028	170
Student-directed Emotional Labor	3.84 (1.04)	4.00 (.79)	3.63 (1.27)	4.00 (1.03)	2.58	(2,168)	.078	.030	170
Social Statuses
White Cisgender Man	.41 (.48)	.38 (.49)	.39 (.50)	.44 (.50)	0.27	(2,171)	.763	.003	173
Heterosexual	.87 (.32)	.86 (.35)	.90 (.30)	.90 (.33)	0.43	(2,165)	.650	.005	167
U.S. Born	.81 (.38)	.81 (.39)	.81 (.40)	.82 (.38)	0.18	(2,170)	.982	.000	172
Professional Statuses
Tenured	.80 (.40)	.79 (.41)	.71 (.46)	.85 (.36)	1.36	(2,179)	.260	.015	181
Full Professor	.43 (.50)	.41 (.49)	.26 (.44)	.55 (.50)	4.34	(2,179)	.014	.046	181
Pre-Professional Department	.11 (.31)	.13 (.33)	.03 (.17)	.12 (.33)	1.34	(2,179)	.265	.150	181
Natural Science or Mathematics Dept	.37 (.48)	.34 (.48)	.43 (.50)	.36 (.48)	0.44	(2,179)	.647	.005	181
Social Science Department	.15 (.35)	.11 (.32)	.17 (.38)	.16 (.37)	0.54	(2,179)	.586	.006	181
Mediators
Student Demands	5.67 (4.76)	6.92 (3.95)	7.24 (4.61)	3.80 (4.89)	10.63	(2,170)	< .001	.111	172
Teaching Resources	3.13 (.98)	2.97 (1.07)	2.82 (.82)	3.47 (.85)	7.27	(2,178)	< .001	.076	180

FWC former women’s college, CC co-ed college, FMC former men’s college

The three colleges have similar numbers of tenure-line faculty (165 to 217) and undergraduate students (2100–2600). None has more than a small number of graduate students (less than 30). Domestic BIPOC students make up 26% of students at FWC, 24% at FMC, and 15% at CC. International students make up between seven and 11% of the students. Between one fifth and one sixth of students are eligible for Pell grants at these colleges where combined costs of tuition, room, and board exceed $70,000 per year. Compared to most predominantly undergraduate institutions, these three have much higher research expectations for faculty as well as lower teaching loads (the equivalent of 18 semester hours per year with small class sizes). These SLACs market themselves to prospective students and their families as providing high levels of interaction with faculty inside and outside the classroom.

All three colleges were completely closed to in-person activities beginning the later part of March 2020, which was the middle of the semester for FWC and CC, but the start of a new completely online trimester for FMC. All three provided workshops on how to teach online. Both CC and FMC administered student ratings of instruction, while FWC did not. CC gave faculty the choice to exclude them from tenure and promotion portfolios while FMC used them strictly for diagnostic rather than evaluative purposes. All three allowed pre-tenure faculty to extend their tenure clocks.

Participants and Procedure

A total of 204 full-time tenured and tenure-track professors participated in the study. For this analysis, we excluded 20 faculty who did not have any teaching responsibilities during spring 2020, retaining three respondents who were not teaching courses but were supervising independent studies, theses, and the like. We also excluded two cases that provided responses to only two items used in this analysis. The final sample included 182 participants (Mage = 50; Myear hired = 2004), comparable in age to the average (49) for tenure-stream faculty in the United States (McChesney & Bichsel, 2020). The vast majority (85%) of the participants lived with another adult at least part-time, and over half (59%) had at least one child 18 or under living at home.

Using a confidential, web-based survey, we collected qualitative and quantitative data from May to September 2020. We sent email invitations to participate to all full-time tenured or tenure-track faculty at the three SLACs. The overall response rate was 37% (See Supplement A in the online supplement for a breakdown of the response rate by rank and SLAC). All respondents provided informed consent and were given the opportunity to participate in a raffle for an iPad. The study received Institutional Review Board approval prior to data collection.

Measures

Emotional Labor

Participants completed measures of emotional labor used by El-Alayli et al. (2018). One measure assessed self-directed emotional labor (4 items, i.e., “I am required to be ‘artificially friendly’ to students;” “I cover or manage my own feelings so as to appear pleasant to my students;” “I am unable to express my true feelings to my students;” “I feel that I have to be nice to students no matter how they treat me”), and one measure assessed student-directed emotional labor (2 items, i.e., “I spend a lot of time helping students feel better about themselves” and “I spend a lot of time helping students deal with stresses and difficulties”). Participants responded on a 5-point scale from 1 (strongly disagree) to 5 (strongly agree). We added a neither agree nor disagree response at the midpoint (3) of the original 4-point response scale used by El-Alayli et al. We averaged the items to create composite scores of self-directed emotional labor (α = .80) and student-directed emotional labor (α = .88), such that higher scores indicate more emotional labor. The Cronbach’s alphas denoted that these items produced the same two highly reliable emotional labor indexes as they did in El-Alayli et al.’s study of faculty from a random sample of colleges and universities in the United States.

Predictor Variables

Social Status

We took an intersectional approach by creating four race by gender groups. The first group consisted of white cisgender men (41% of participants). To avoid eliminating the two respondents who identified as non-binary, both of whom were white, we combined them with white cisgender women to generate a group that shared a privileged racial identity but a lower status gender identity (42%). Only 17% of the sample identified as BIPOC, too few to separate into specific racial or ethnic groups. Therefore, we made only two more intersectional groups: BIPOC cisgender men (6%) and BIPOC cisgender women (11%). We analyzed two other social statuses: sexual orientation and nativity. Sexual orientation also had too few participants in any category other than heterosexual/straight. Therefore, we collapsed the other responses into an LGBTQ + category. We coded heterosexual/straight faculty as 1 and LGBTQ + as 0. Only 19% of the sample were born outside of the United States, so we grouped all faculty born outside of the U.S. and coded them as 0 and coded all U.S. born individuals as 1.

Professional Status

Following Tunguz’s (2016) approach, the first professional status variable measured tenure (we coded pre-tenured individuals as 0 and tenured as 1). We build on her work by including a dummy variable measuring rank (we coded assistant and associate professors as 0 and full professors as 1). Forty-three percent of the participants held the rank of full professor, 36% associate professor, and 21% assistant professor. Next, we grouped departments into broad disciplinary categories. Using the most highly feminized fields—the arts and humanities departments (including history)—as the referent category, we created dummy variables to measure three more masculine disciplinary areas: natural science and mathematics departments (including psychology), social science, and pre-professional departments (i.e., education, engineering, management and business, and social work). While education and social work are female-identified fields, they had very few faculty members. Therefore, the engineering and management and business departments dominated the pre-professional category. Finally, to measure the effect of male-dominated institutions, we constructed two dummy variables, one for FMC (37%) and the other for CC (19%), using FWC (44%) as the referent category.

Mediating Variables

Student Demands

We adapted five student demand items from El-Alayli et al. (2018), who had separated them into three categories: standard work demands, solicitation of special favors, and friendship behaviors. We used one of the original work demand items (i.e., “students would send emails with questions about course materials and assignments”) and one friendship behavior item (i.e., “students would discuss personal issues such as mental health with you”) because the other items in these categories involved in-person contact that was no longer possible in spring 2020. We adapted three of original items from the solicitation of special favors category: “students would ask for adjustments or exceptions on grading (higher grades, extra credit, resubmissions);” “students would ask to meet (either in-person or via phone/Internet) and expect to be able to meet right away without scheduling an appointment;” and “students would ask for adjustments or exceptions on assignments (extensions, exemptions, reductions).” To capture the potential increase in the types of behaviors that El-Alayli et al. (2018) labeled special favors, we added five items to measure demands that might have increased due to the pandemic, including one friendship behavior (i.e., “you exchanged text messages with students”) and four special favors (i.e., “students requested that class be cancelled for a day;” “students discussed with you whether they should take your course Pass/Fail;” “students discussed issues with Wi-Fi, computer access, or phone access that kept them from completing work for your course;” and “you needed to go to extra lengths to get in touch with students who dropped out of contact [i.e., sending extra emails, contacting student advising office, etc.]”).

Rather than ask participants to estimate how often students engaged in these behaviors in the typical term as El-Alayli et al. (2018) did, we asked them to rate how frequently these ten experiences occurred during the spring 2020 term compared to fall 2019 on a 5-point scale ranging from -2 (much less often) to 2 (much more often). We dropped two items, exchanging text messages with students and being asked to cancel class, from the composite index because so few participants reported they occurred (73 and 49 respectively). We summed the eight remaining items (one standard work demand, one friendship behavior, and six special favors) to form a single student demands index with the same high level of reliability (α = .77) that El-Alayli et al. found for their special favor index. Scores ranged from –13 to 15. Higher scores indicate that faculty experienced more student demands more often during spring 2020 than pre-pandemic.

Teaching Resources

We used three items to assess participants’ teaching resources during the pandemic. Two of these items assessed participants’ ability to dedicate adequate time and space to teaching (i.e., “I have been able to dedicate enough time to my teaching,” “I have been able to dedicate enough space to my teaching”) and one item assessed participants’ ability to concentrate on their teaching (i.e., “I have had trouble concentrating on my teaching”). Participants rated their responses on a five-point Likert scale from 1 (strongly agree) to 5 (strongly disagree). After reverse-coding the adequate time and space variables, the three items were averaged into a single index measuring teaching resources with high reliability (α = .75). Higher scores indicate more teaching resources.

Analysis Plan

The analysis proceeded in three stages. First, we explored how the variables differed among the three SLACs by conducting a one-way ANOVA on the original dataset because SPSS does not calculate pooled F-ratios for ANOVAs (van Ginkel & Kroonenberg, 2014). We then examined how the four intersectional race and gender statuses were associated with both self- and student-directed emotional labor by conducting a one-way MANOVA on the original dataset. Next, we used OLS regression on the imputed dataset first to assess whether the social and professional status predictors were significantly related to the two mediators and then to model the relationship between the two types of emotional labor and social and professional statuses, first separately and then simultaneously. Finally, we added the mediators one at a time in separate regression models before putting both in the final model.

We performed collinearity diagnostics for each regression model on the original dataset as well as the 20 imputed datasets. No VIF value was greater than 1.60, indicating that there was no multicollinearity. To examine the statistical significance of any mediation, we conducted Sobel tests using the Aroian version (Baron & Kenny, 1986; Sobel, 1982) (see Supplement B in the online supplement for more detailed information on the Aroian version). We replicated the analysis on the subset of faculty with complete data across all variables in this analysis (n = 165). The results were consistent with those from the imputed pooled data presented below.

Results

Preliminary Analyses

Analysis of patterns of missing data for the variables used in this study and the 182 participants included in this sample indicated that 89% of cases had no missing data, 27% of the variables were not missing data for any respondent, and no variable had more than 8% (sexual orientation) of its values missing. Little’s Missing Completely at Random test was not significant, (40, N = 182) = 46.39, p = .23, indicating that these data were missing at random.

We used multiple imputation to preserve sample size. The imputation model included all the variables used in this analysis as well as variables from the larger dataset that may have improved the imputation (i.e., year hired, did the faculty member have to switch to remote instruction during the spring 2020 term, research resources, scholarly productivity, work strain, English as first language, living with partner, and having children 18 or under living at home). Twenty datasets were imputed. We present the pooled models and statistics when available.

Race, Gender, and Emotional Labor

As Table 1 indicates, faculty reported engaging in more student-directed than self-directed emotional labor in spring 2020 compared to fall 2019. When asked about performing student-directed emotional labor, two-thirds (66%) of faculty agreed or strongly agreed that they spent a lot of time helping students feel better about themselves, and nearly three-quarters (73%) of faculty indicated that they spent a lot of time helping students deal with stresses and difficulties. Levels of agreement with the self-directed emotional labor items were lower with 51% agreeing or strongly agreeing that they covered or managed their own feelings to appear pleasant to their students, 40% indicated they had to be nice to students regardless of how they were treated, 38% reported they were unable to share true feelings with students, and only 21% stated that they had to be “artificially friendly” to students. There were no significant mean differences among the SLACs in the two forms of emotional labor, but faculty at FMC reported fewer student demands and more teaching resources than the faculty at CC and FWC.

While a visual assessment of the boxplots showed no outliers, Shapiro Wilk’s test (p < .05) revealed that the student-directed emotional labor data was not normally distributed for any of the race and gender groups. Self-directed emotional labor was normally distributed for BIPOC cisgender men, BIPOC cisgender women, and white cisgender women and gender non-conforming (GNC) faculty, but not for white cisgender men. Levene’s Test of Homogeneity of Variance revealed homogeneity of variances for self-directed emotional labor (p > .05), but not for student-directed emotional labor (p < .05). To account for the strongly negatively skewed student-directed emotional labor data, we applied a reflect and logarithmic transformation to the original variable (Laerd Statistics, 2015). While Shapiro Wilk’s test (p < .05) indicated the transformed student-directed emotional labor variable was still not normally distributed, Levene’s Test of Homogeneity of Variance revealed homogeneity of variance (p > .05). Though it still violated the assumption of normality, we proceeded with the one-way MANOVA because the assumption of homogeneity of variance was now met and this test is robust to deviations of normality (Laerd Statistics, 2015). Furthermore, Pearson correlation (r = –.372, p ≤ .001) indicated there was no multicollinearity, and Mahalanobis distance (p > .001) determined there were no multivariate outliers. In the text of the article, we present the means and standard deviations of the transformed student-directed emotional labor variable as well as the original variable. See Supplement C in the online supplement for the original means and standard deviations for the student-directed emotional labor variable prior to transformation.

The number of participants in the four intersectional groups violated the assumption of equal sample sizes with substantially more white participants (71 white cisgender women or GNC and 68 white cisgender men) than BIPOC (20 BIPOC cisgender women and 10 BIPOC cisgender men), and twice as many BIPOC women as BIPOC men. Despite the large difference in group sizes, Box’s test of equality of covariance matrices indicated that there was homogeneity of variance–covariance (p = .63).

The one-way MANOVA indicated no significant difference between intersectional race and gender identity on the combined dependent variables, F(6, 328) = 4.90, p = .08; Wilk’s Λ = .84; partial η2 = .08. Separate univariate ANOVAs revealed a significant effect of intersectional identity on self-directed emotional labor (F(3, 165) = 3.77, p = .01; partial η2 = .06). A Tukey post hoc test showed no significant differences between white men (M = 2.61, SD = 1.06) and white cisgender women and GNC faculty (M = 3.06, SD = 0.96, p = .06, d = .45), BIPOC men (M = 3.50, SD = 1.26, p = .06, d = .76) nor BIPOC women (M = 3.17, SD = 1.20, p = .16, d = .49).

A significant effect of intersectional identity on student-directed emotional labor was also revealed (F(3, 165) = 9.01, p ≤ .001, partial η2 = .14). A Tukey post hoc test indicated that white men reported less student-directed emotional labor (M = .37, SD = .20; M = 3.37, SD = 1.20) than white cisgender women and GNC faculty (M = .22, SD = .18, p ≤ .001, d = .79; M = 4.18, SD = .83), BIPOC men (M = .19, SD = .14, p = .02, d = 1.04; M = 4.40, SD = .46), and BIPOC women (M = .24, SD = .19, p = .02, d = .67; M = 4.13, SD = .74).

In short, white cisgender men reported engaging in less student-directed emotional labor than the three other intersectional groups. None of the other race and gender groups were significantly different from each other on either type of emotional labor. Therefore, in the following multivariate analyses, we present an analysis of the relationship between white cisgender men and both forms of emotional labor using the other three intersectional groups as a single referent category.

Race, Gender, Student Demands, and Teaching Resources

We tested whether the two mediators (student demands and teaching resources) varied by intersectional race and gender identities as well as the other social and professional statuses (see Table 2). Consistent with the results in Table 1, faculty at FMC reported fewer student demands (p < .001) and more teaching resources (p = .01) than the faculty at the other two SLACs even after controlling for the other professional statuses, particularly rank, and the other social statuses, particular race and gender. Consistent with expectations, white cisgender men reported fewer student demands (p = .01) and more teaching resources (p = .03) than faculty with other intersectional identities.

Table 2

Imputed OLS regression of mediators on social and professional status (N = 182)

	Student Demands	Teaching Resources
	b(SE)	b(SE)
White Cisgender Man	–2.197** (.746)	.341* (.154)
Heterosexual	.949 (1.131)	–.063 (.232)
US Born	.334 (.900)	–.262 (.187)
Tenured	–.445 (.949)	.144 (.197)
Full Professor	–1.334 (.801)	.324* (.163)
Pre-Professional Dept	–1.662 (1.275)	–.040 (.239)
Natural Science Dept	–.295 (.795)	–.009 (.165)
Social Science Dept	.228 (1.056)	.202 (.220)
FMC	–2.665*** (.749)	.419** (.154)
CC	.211 (.929)	–.113 (.191)
Constant	7.615*** (1.387)	2.842*** (.286)
Adj R¯^2	.16***	.12***

*p < .05; **p < .01; ***p < .001

Multivariate Analysis of Emotional Labor

Self-Directed Emotional Labor

The first three hypotheses were that faculty with more privileged social statuses performed less self- and student-directed emotional labor than those with less privileged ones. As the first column of Table 3 (Model 1) presents, being a white cisgender man was negatively associated with self-directed emotional labor (p = .01). Neither of the other two social statuses, sexual orientation (p = .35) or nativity (p = .43), had significant relationships with self-directed emotional labor.

Table 3

Imputed OLS unstandardized regression coefficients predicting self-directed emotional labor (N = 182)

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
	b(SE)	b(SE)	b(SE)	b(SE)	b(SE)	b(SE)
Social Statuses
White Cisgender Man	–.510** (.167)		–.412* (.170)	–.275 (.167)	–.307 (.166)	–.229 (.165)
Heterosexual	.252 (.269)		.251 (.283)	.191 (–.277)	.231 (.273)	.192 (.271)
US Born	–.165 (.211)		–.165 (.209)	–.185 (.202)	–.246 (.202)	–.243 (.198)
Professional Statuses
Tenured		–.080 (.217)	–.052 (.224)	–.024 (.216)	–.007 (.214)	–.003 (.211)
Full Professor		–.439* (.182)	–.323 (.187)	–.240 (.182)	–.223 (.185)	–.184 (.181)
Pre-Professional Dept		–.452 (.282)	–.504 (.280)	–.401 (.275)	–.516* (.270)	–.437 (.269)
Natural Science Dept		–.074 (.179)	–.106 (.182)	–.088 (.177)	–.109 (.176)	–.095 (.173)
Social Science Dept		.408 (.243)	.365 (.244)	.351 (.236)	.427 (.236)	.403 (.233)
FMC		–.231 (.176)	–.229 (.174)	–.063 (.175)	–.100 (.172)	–.006 (.173)
CC		.071 (.217)	.085 (.214)	.071 (.206)	.049 (.206)	.048 (.202)
Mediators
Student Demands				.062*** (.018)		.046* (.018)
Teaching Resources					–.309*** (.084)	–.238** (.086)
Constant	3.033*** (.311)	3.251*** (.210)	3.285*** (.326)	2.810*** (.347)	4.164*** (.384)	3.609*** (.436)
Adj R¯^2	.048*	.073*	.101**	.162***	.167***	.195***

*p < .05; **p < .01; ***p < .001

Turning to hypotheses 4 and 5, that tenured faculty and full professors would perform less self- and student-directed emotional labor than untenured faculty and assistant and associate professors, Model 2 in Table 3 indicates that being tenured (p = .71) was not significantly negatively associated with self-directed emotional labor, but that being a full professor was (p = .02). The sixth hypothesis was that faculty in more male-identified disciplines would engage in less self- and student-directed emotional labor than faculty in more female-identified disciplines, and the seventh was that faculty teaching at the more male-identified colleges (FMC and CC) would engage in less self- and student-directed emotional labor than faculty at the more female-identified college (FWC). Neither discipline (pre-professional departments p = .11, natural science p = .68, social science p = .09) nor college of employment (FMC p = .19 and CC p = .74) were significantly associated with self-directed emotional labor.

We then regressed self-directed emotional labor on all the social and professional status variables simultaneously. Model 3 in Table 3 indicates that while being a white cisgender man was still significantly associated with lower levels of self-directed emotional labor, being a full professor was no longer significant (p = .02 and p = .08 respectively).

In the next step, we asked whether the mediators explained the relationship between intersectional race and gender identity and the two types of emotional labor. As Model 4 indicates, the student demands for special favors index was positively associated with self-directed emotional labor (p ≤ .001). The Aroian version of the Sobel test (z = –2.186, SE = .062, p = .03) confirmed that student demands fully mediated the relationship the professor’s race and gender had with self-directed emotional labor. The white cisgender man dummy variable was no longer significant after controlling for student demands (p = .10). As Model 5 indicates, the teaching resources index was negatively associated with self-directed emotional labor (p ≤ .001). The relationship between the intersectional race and gender variable and self-directed emotional labor was no longer significant (p = .06), nor was the Aroian test (z = –1.848, SE = .057, p = .06). When both mediators were included in the final regression model (Model 6) along with the social and professional status variables, the significant positive relationship between student demands (p = .01) and self-directed emotional labor and the significant negative relationship between teaching resources (p = .01) and self-directed emotional labor remained. All the other variables, including the intersectional race and gender one (p = .16), were no longer significantly associated with self-directed emotional labor.

Student-Directed Emotional Labor

As the first column of Table 4 (Model 1) presents, being a white cisgender man was also negatively associated with student-directed emotional labor (p ≤ .001). Neither sexual orientation (p = .25) nor nativity (p = .76) had significant relationships with student-directed emotional labor. Model 2 in Table 4 indicates that being tenured was not significantly negatively associated with student-directed emotional labor (p = .26), but that being a full professor was (p = .01). Consistent with the sixth hypothesis, working in a pre-professional department as well as in a natural science or mathematics department (including psychology) were negatively associated with student-directed emotional labor (p ≤ .001 and p = .04, respectively). Testing the seventh hypothesis revealed that there was no statistically significant difference between FWC and the two male-identified colleges: FMC (p = .07) and CC (p = .44). Model 3 in Table 4 indicates that when regressing student-directed emotional labor on the social and the professional status variables simultaneously, all the variables that were significant in models 1 and 2 remained significant except for working in a natural science department (p = .10).

Table 4

Imputed OLS unstandardized regression coefficients predicting student-directed emotional labor (N = 182)

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
	b(SE)	b(SE)	b(SE)	b(SE)	b(SE)	b(SE)
Social Statuses
White Cisgender Man	–.804*** (.157)		–.698*** (.153)	–.512*** (.145)	–.666*** (.154)	–.520*** (.145)
Heterosexual	–.274 (.239)		–.195 (.241)	–.276 (.218)	–.200 (.240)	–.276 (.218)
US Born	–.059 (.197)		.009 (.188)	.019 (.172)	.015 (.189)	.009 (.173)
Professional Statuses
Tenured		–.226 (.200)	–.119 (.196)	–.081 (.179)	–.105 (.196)	–.085 (.179)
Full Professor		–.489** (.167)	–.328* (.163)	–.215 (.151)	–.297 (.166)	–.225 (.153)
Pre-Professional Dept		–1.116*** (.262)	–1.103*** (.250)	–.964*** (.230)	–1.107*** (.250)	–.958*** (.231)
Natural Science Dept		–.348* (.172)	–.272 (.166)	–.247 (.152)	–.273 (.166)	–.246 (.153)
Social Science Dept		.063 (.232)	.125 (.224)	.106 (.204)	.144 (.224)	.097 (.206)
FMC		–.296 (.164)	–.280 (.155)	–.055 (.148)	–.240 (.159)	–.065 (.149)
CC		–.157 (.201)	–.118 (.190)	–.136 (.173)	–.129 (.189)	–.132 (.173)
Mediators
Student Demands				.084*** (.016)		.087*** (.016)
Teaching Resources					–.094 (.078)	–.040 (.074)
Constant	4.355*** (.283)	4.611*** (.198)	4.851*** (.288)	4.209*** (.292)	5.119*** (.359)	4.074*** (.382)
Adj R¯^2	.141***	.171***	.266***	.388***	.269***	.385***

*p < .05; **p < .01; ***p < .001

We asked next whether the mediators explained these relationships between social and professional statuses and student-directed emotional labor. As Model 4 in Table 4 indicates, the student demands index was positively associated with student-directed emotional labor (p ≤ .001). It partially mediated the effects of a professor’s race and gender with a reduction of almost two-tenths of a point in the unstandardized regression coefficient (z = –2.533, SE = .073, p = .01). In contrast, the teaching resources index (p = .23) was not significantly related to student-directed emotional labor (see Model 5). It had no effect on the relationship between race and gender identity and student-directed emotional labor (z = –.984, SE = .033, p = .32). When we included both mediators in the final model (Model 6) along with the social and professional status variables, the positive relationship between student demands and student-directed emotional labor and the negative relationships between the professor’s race and gender as well as working in a pre-professional department and student-directed emotional labor remained.

Discussion

This study surveyed 182 full-time tenured and tenure-track faculty from three small private liberal arts colleges to examine the effect of social and professional statuses on emotional labor during the emergency switch to remote instruction in spring 2020. We found that faculty in pre-professional departments engaged in less student-directed emotional labor than faculty in liberal arts and sciences departments. Consistent with previous research (e.g., El-Alayli et al., 2018), men reported significantly less emotional labor than women, but when we examined race and gender intersectionally, we specifically found that white cisgender men reported significantly less emotional labor than BIPOC cisgender men, BIPOC cisgender women, and white cisgender women and GNC faculty.

Student demands mediated the relationship between race and gender status and emotional labor. Extending past research (El-Alayli et al., 2018; O’Meara et al., 2017) showing that women faculty received more extra work requests from students than men, we found that white cisgender men reported fewer student demands than the three other intersectional race and gender groups. Intersectional race and gender identity significantly predicted students’ special favor requests, which in turn significantly predicted both self-directed and student-directed emotional labor. The mediation analysis revealed that the initial relationship between intersectional race and gender identity and self-directed emotional labor was a function of the lower number of student demands for special favors that white cisgender men received compared to faculty with other race and gender identities. Similarly, white cisgender men appeared to have engaged in less student-directed emotional labor than their colleagues, in part because students requested fewer special favors from them.

Students making more demands of faculty who were not white cisgender men placed greater emotional labor burdens on them, leading to more time-consuming and emotionally difficult interactions with students. O’Meara et al. (2017) found that students and former students were the most likely sources of requests for faculty to engage in additional work. As a result, women had to “consider and come up with more responses” to these student demands, which required them to do more emotional labor to manage their own emotional response to the student request as well as the student’s emotional response to their answer to that request (O’Meara et al., 2017, p. 1179). As a result of students making more demands for special favors of white women, BIPOC women, and BIPOC men, the job of being a faculty member was different for them than for white men. In short, this study revealed that the extra self-directed emotional labor and at least some of the student-directed emotional labor experienced by BIPOC men, BIPOC women, and white cisgender women and GNC faculty during the COVID-19 pandemic in spring 2020 was related to more student requests for special favors.

The status shield afforded white cisgender men by their race and gender appeared to have protected them from student requests for special favors. Lacking the status shield white cisgender men have, other faculty received more student demands in the early stages of the pandemic. Students may still have seen women as academic moms (Bernard, 1964) who they expected to be more approachable and more nurturing when they needed help (El-Alayli et al., 2018).

Similarly, the academic culture in pre-professional engineering as well as management and business departments may have been sufficiently masculinized to create a barrier that students did not want to breach by asking for special favors even during the early stages of the COVID-19 pandemic (Hall & Sandler, 1982; Lee & McCabe, 2021). White cisgender men in pre-professional departments made the least investment of student-directed emotional labor even after controlling for significant differences in student demands (and teaching resources). The results suggested that faculty with the most privileged social statuses (white cisgender men) in the most masculinized disciplines (pre-professional) had a status shield that enabled them to avoid engaging in as much emotional labor as faculty with less privileged race and gender identities and in less masculinized liberal arts and sciences disciplines.

Limitations and Future Research Directions

This study contains several limitations that should be considered. First, this study was cross-sectional and the data collection took place in the beginning of the COVID-19 pandemic (May through September 2020). As such, these data only captured the immediate effects of the pandemic on faculty emotional labor. In addition, some faculty completed the survey toward the end of the spring term, some immediately after their spring term had ended, while others completed it during the summer. Their answers may have varied based on how much time had elapsed, if any, since the spring 2020 term ended.

Furthermore, because we collected these data at a time when faculty were still struggling to adapt to remote learning as well as the additional personal and professional challenges caused by the pandemic, the sample was likely to be biased in a number of ways. For example, it is possible that the faculty who responded to this survey were those who were feeling less overwhelmed and therefore more capable of taking the time to participate. Conversely, it is also possible that faculty who were feeling extremely stressed were more likely to complete the survey because it gave them an outlet to describe what they were experiencing.

Additional limitations include the small sample size as well as the geographical location (New York) and type of school (SLACs) surveyed. Overall, the emotional labor demands on the faculty in this sample may have been substantially higher than those for faculty at larger institutions or research-intensive institutions. A larger sample size might have provided enough power to detect additional statistically significant effects, such as the lower level of self-directed emotional labor at FMC compared to the other two SLACs. Furthermore, a larger sample from a broader range of institutions would have enabled us to ascertain whether white women, BIPOC women, and BIPOC men experienced even higher levels of emotional labor in institutions or departments where their numbers were few.

Finally, as a function of the geographical location and type of schools surveyed, as well as general disparities in BIPOC representation among faculty nationwide, there were very few BIPOC faculty in the sample. The difference in means on self-directed emotional labor between BIPOC women and white men may have reached statistical significance if there had been a larger number of BIPOC women in the sample. Future research with larger sample sizes for BIPOC men and women is needed to determine whether there are statistically significant differences in the amount of emotional labor BIPOC women were doing during the COVID-19 pandemic compared to white women and BIPOC men in addition to white men. Previous research has demonstrated that students and colleagues expect BIPOC faculty, especially BIPOC women, to invest more emotional labor in BIPOC students and do more diversity, equity, and inclusion service work (Clark et al., 2020; Hirshfield & Joseph, 2012; Moore et al., 2010; Oleschuk, 2020; Settles & Linderman, 2020; Turner, 2002). It is likely, therefore, that the emotional labor demands for BIPOC faculty were even higher than these data suggest. Emotional labor demands for BIPOC faculty were also likely higher at predominantly white institutions, such as the three surveyed in this study, than they were at predominantly Black institutions. It may be, however, that the emotional labor demands on BIPOC women were already so high pre-pandemic that they did not experience the same increase that white women and BIPOC men did.

Future research should implement probabilistic sampling techniques to generate a nationally representative sample of higher education institutions and the faculty employed by them. Doing so would allow for greater generalizability of findings and a broader representation of faculty identities and experiences. Longitudinal research would also be useful to understand long-term career outcomes of these differences in emotional labor. Moreover, future research should consider the impact that course content or course level has on emotional labor, as some classes may have been more emotionally laborious than others.

Practice Implications

Given that the content of the job of professor differed depending on the incumbent’s intersectional race and gender identity, administrators should consider the varying amounts of emotional labor that faculty engaged in to meet the needs of students when evaluating and rewarding the work of faculty during the pandemic. Administrations should ask faculty to document the emotional labor that they performed for students during the pandemic on their annual reports and in their tenure and promotion portfolios so it can be recognized and more importantly rewarded, especially in high stakes personnel decisions (Gonzales & Griffin, 2020; King & Frederickson, 2021; Misra et al., 2020).

Perhaps more importantly, to ensure equity, personnel policies and practices need to be modified to consider the differential workload generated by emotional labour (for best practices, see Malisch et al., 2020; Misra et al., 2020; Oleschuk, 2020; Settles & Linderman, 2020). King and Frederickson (2021) recommended that to “account for the greater teaching demands of moving courses online, institutions should consider shifting their percentage balances (for the relative importance of research, teaching, and service) for faculty evaluation for the duration of the pandemic” (p. 23). Rather than simply stopping the tenure clock, which delays access to job security, a promotion, and a raise, colleges and universities should recalibrate the criteria used in personnel decisions to assess achievement relative to opportunity (Carpenter et al., 2021; Hill et al., 2014; Misra et al., 2021). Alternatively, colleges and universities could work to limit students’ demands on faculty, particularly those that are outside of their academic areas of expertise, by providing more support services, especially mental health services.

Conclusion

The results of this study suggest that the performance of student-directed emotional labor varied by social and professional statuses, specifically intersectional race and gender as well as discipline. Consistent with prior literature on faculty emotional labor (El-Alayli et al., 2018; Mahoney et al., 2011; Tunguz, 2016), we found that in the wake of pandemic disruptions white cisgender women and GNC faculty as well as BIPOC men and women performed more emotional labor than their white cisgender men colleagues, reducing the time and energy that they could expend on scholarship and other work demands.

In their study of male nurses, Cottingham et al. (2015) concluded that men “take their privileged status to work, even in work stereotyped as feminine,” providing them with a status shield (p. 387). Conversely, we found that women took their disadvantaged status into college teaching, a traditionally male-dominated profession. Women carried a weaker status shield against student requests for special favors during the pandemic than their white male colleagues. As men and women have moved into non-traditional occupations, their gender has followed them, creating higher demands for emotional labor from workers with marginalized racial and gender identities, and reproducing racial and gender inequality within these occupations.

Below is the link to the electronic supplementary material.

Supplementary file1 (DOCX 15 KB)