The status of race in public sector work: Implications for emotion management and job satisfaction

ABSTRACT

African Americans consistently report lower levels of job satisfaction relative to whites. Using survey data from 1,456 public service employees, we examine whether racial disparities in job satisfaction are related to how African Americans and whites manage their emotions while at work. We contend that race acts as a master status within the workplace that locates African Americans in a subordinate social position to whites and may contribute to greater emotion management effort and greater work-related consequences. The results indicate that, together with traditional indicators of job satisfaction, extensive emotion management efforts of African Americans explain their lower levels of job satisfaction relative to whites.

Notes

¹ The emotional proletariat are similar to “reproductive laborers,” engaged in work that is of low status but necessary for the maintenance of the labor force (e.g., food preparation workers, dishwashers, counter attendants, janitors, housekeepers, laundry workers) (Duffy 2005). Workers in these jobs can be contrasted with “organized emotional care” workers to which organizations provide the job flexibility and support necessary to engage emotionally in the socio-relational aspects of the job (Lopez 2006; Rakovski and Price-Glynn 2010).

² The first author collected the data while living in Tennessee. The state is Tennessee’s largest employer. In 2004, Tennessee was similar to other Southern states on key economic indicators including GDP and unemployment rate (U.S. Bureau of Labor Statistics 2004).

³ The sampling frame included all workers who have been in their position for at least six months and were working in career service occupations in the areas administrative, office, or clerical work; the natural sciences and health care; information systems or telecommunications; public safety; maintenance; and social, human, or employment services. Persons in “executive service” positions who served the governor directly were not included in the sampling frame.

⁴ We employed multiple imputation, using the multivariate normal model and mi command in Stata (M = 20) to impute missing data on all variables (Allison 2002; Royston 2004). The regression coefficients presented in the tables are estimates that have been averaged across the imputed datasets, and standard errors were estimated following Rubin (1987). The Adjusted R² values were estimated using the mibeta command in Stata, written by Yula Marchenko (see Harel 2009). Multiple imputation produced the same pattern of results as listwise deletion of missing data in terms of the magnitude and significance of associations among race, emotion management, and job satisfaction.

⁵ Emotion management may take many forms including those commonly referred to as surface acting or deep acting. Surface acting involves displaying an appropriate emotion but not altering the internally felt emotion (Hochschild 1983). An alternative strategy, deep acting, involves working to alter the internal feeling itself (e.g., trying not to feel an emotion or trying to feel the emotion that should be expressed). Because we are interested in the general act of controlling emotions, we use a measure of the extent to which workers manage internally felt emotions.We also ran the analyses reported in this paper using a variable that measured Surface Acting in particular in place of Emotion Management Effort. The results were substantively similar to those using Emotion Management Effort and are available upon request. African Americans engaged in significantly more surface acting than whites, and surface acting was associated with decreased job satisfaction.

⁶ We also created alternative measures of Job Satisfaction, the Requirement of Emotion Management, and Emotion Management Effort by dichotomizing the Likert scale responses into agree and disagree categories and summing those items. The results using these alternative measures replicate the results presented here and are available upon request.

⁷ Although these items have been used in previous research (e.g., Erickson and Ritter 2001; Sloan 2014), their internal consistency scores (Cronbach’s alpha = .64 and .66, respectively) are near the lower limit of acceptability for scale measures (Hair, Black, and Babin 2010), indicating a limitation of the data. Treated separately, each scale item is significantly correlated with both Job Satisfaction and Emotion Management Effort.

⁸ Supervising Others was measured on the questionnaire in five ordinal categories (1 = 1–3, 2 = 4–6, 3 = 7–10, 4 = 11–19, 5 = 20 or more); Number of Supervisors was also measured in five ordinal categories (1 = 1, 2 = 2, 3 = 3, 4 = 4–5, 5 = 6 or more, with a separate response for 0 supervisors). After reverse coding the number of supervisors variable, assigning a value of 6 to a response of 0 supervisors, and adding the supervising others variable, the workplace status measure had a range of 1–11. For example, a respondent who does not report to a supervisor and supervises more than 20 other workers (a value of 5 in the ordinal response set) received a score of 11 (the highest score), and a respondent who reports to six or more people and supervises no one received a score of 1. In addition to using the Workplace Status measure, we ran all analyses with the original two ordinal variables included separately. Those analyses produced the same patterns of results reported here and are available by request.

⁹ We also ran our analysis with a control measure of the experience of agitated emotions in the workplace (e.g., Erickson and Ritter 2001). Controlling for agitated emotional experiences did not significantly change the relationships among emotion management effort, race, and job satisfaction. The analyses are available by request.

¹⁰ Because Requirement of Emotion Management is an ordinal measure, we also tested Hypothesis 1 using ordered logistic regression. Results were consistent with the OLS analysis. For consistency and ease of interpretation, we report the OLS results. Results of the ordinal analyses are available by request.

¹¹ We analyzed nonresponse by gender and by region of the state; no response bias was revealed. T tests comparing the sample to the state population as recorded in the 2000 Census showed that the sample contains significantly more non-whites, women, and married individuals than the proportions in the state as a whole (U.S. Census Bureau 2000). This overrepresentation of women and racial minorities is consistent with the demographic composition of the public sector workforce.

Additional information

Notes on contributors

Melissa M. Sloan

Melissa M. Sloan is Associate Professor of Sociology and Interdisciplinary Social Sciences at the University of South Florida Sarasota-Manatee. Her research interests include social psychology, emotion, and mental health/illness. Her recent work examines the relationships between social statuses, emotional behaviors in the workplace, and worker wellbeing.

James D. Unnever

James D. Unnever is Professor of Criminology at the University of South Florida Sarasota-Manatee. He was the Recipient of the Donal A. J. MacNamara Award by the Academy of Criminal Justice Sciences in 2009 for the best article published that year. The author of over 40 publications appearing in such journals as Social Forces, Criminology, Social Problems, Journal of Research in Crime and Delinquency, and Justice Quarterly, he was ranked as the fifth most innovative author in criminology from 2000–2010. His areas of expertise include race and crime, public opinion about crime-related issues, including the death penalty, the testing of theories of crime, and school bullying.