The Ideal Worker Norm and Workplace Social Support among U.S. Workers

ABSTRACT

Research has shown that the ideal worker norm stipulating workers should be completely devoted to their jobs without interference from outside responsibilities creates difficulties for workers. At the same time, scholarship continues to emphasize the positive outcomes associated with coworker and supervisor support in making it easier for workers to combine work and family. Yet we know little about what shapes the extent to which workers have access to supportive coworkers and supervisors. This study brings together these two strands of scholarship to explore the relationships between the perceived presence of the ideal worker norm and two forms of workplace social support: coworker support and supervisor support. Using data from a nationally representative sample of U.S. workers, this study finds that the perceived presence of the ideal worker norm—regardless of gender and largely irrespective of parenthood and elder-care responsibilities—is associated with decreased levels of both forms of support. Taken together, these findings contribute to the literature by documenting the impact of the ideal worker norm on the workplace social support to which workers have access.

Disclosure Statement

No potential conflict of interest was reported by the authors.

Notes

¹ A previous version of this manuscript incorporated four different types of support—general coworker support, family-related coworker support, general supervisor support, and supervisor work-family support—with similar results.

² Although the SMC-FCS (Bartlett et al. 2015) approach to multiple imputation appears very promising for imputation in the presence of interactions and other derived variables, this is an ongoing area of research, and other approaches have been often used in the past. These include (a) imputation of primary variables alone followed by direct calculation of derived values—what Von Hippel (2009) calls “impute, then transform” (IT); (b) van Buuren and Groothuis-Oudshoorn’s (2011) “passive imputation” (PI), where only primary variables are imputed, but the transformations are used to impute the dependent variable; and (c) Von Hippel’s (2009) “transform, then impute” method, where interactions and other derived variables are computed from the incomplete dataset, and then treated as “just another variable” (JAV) for imputation purposes. As a check on the dependency on the imputation method, we also ran the analyses on (sets of) imputed datasets using all these methods and found broadly similar results. The only substantive difference was on the ideal worker X elder-care responsibilities interaction term in supervisor support Model 2, where the coefficient shifted from—.055 (SMC) to—.49 (JAV),—.046 (PI), or—.045 (IT), and mild significance (p = 0.042 SMC) changed to not-quite-significance: p = 0.068 (JAV) or 0.088 (PI and IT). Although the SMC approach is designed specifically to be more accurate with derived variables such as interactions, this might be further reason to take the significance of the interaction term (ideal worker norm X elder-care responsibilities in predicting supervisor support) with a grain of salt.

Additional information

Notes on contributors

Krista Lynn Minnotte

Krista Lynn Minnotte is professor of sociology at the University of North Dakota. Her teaching and research centers on examining the interrelationships among gender, work, and family, with an emphasis on understanding how culture—both the workplace culture and broader societal culture—shape people’s experiences navigating paid work and family. She may be reached via e-mail at [email protected].

Michael C. Minnotte

Krista Lynn Minnotte is professor of sociology at the University of North Dakota. Her teaching and research centers on examining the interrelationships among gender, work, and family, with an emphasis on understanding how culture—both the workplace culture and broader societal culture—shape people’s experiences navigating paid work and family. She may be reached via e-mail at [email protected].

Michael C. Minnotte received his PhD in statistics from Rice University in 1993 and is a professor of mathematics at the University of North Dakota. His statistical work developing nonparametric methods of modeling and feature detection for populations that are not well fit by standard models has appeared in journals including The Annals of Statistics, Journal of the American Statistical Association, and Journal of the Royal Statistical Society. He may be reached at [email protected].