How to Break the Cycle of Low Workforce Diversity: A Model for Change.

Social justice concerns but also perceived business advantage are behind a widespread drive to increase workplace diversity. However, dominance in terms of ethnicity, gender or other aspects of diversity has been resistant to change in many sectors. The different factors which contribute to low dive...

Full description

Bibliographic Details
Main Authors: Katherine R O'Brien, Marten Scheffer, Egbert H van Nes, Romy van der Lee
Format: Article
Language:English
Published: Public Library of Science (PLoS) 2015-01-01
Series:PLoS ONE
Online Access:https://doi.org/10.1371/journal.pone.0133208
Description
Summary:Social justice concerns but also perceived business advantage are behind a widespread drive to increase workplace diversity. However, dominance in terms of ethnicity, gender or other aspects of diversity has been resistant to change in many sectors. The different factors which contribute to low diversity are often hotly contested and difficult to untangle. We propose that many of the barriers to change arise from self-reinforcing feedbacks between low group diversity and inclusivity. Using a dynamic model, we demonstrate how bias in employee appointment and departure can trap organizations in a state with much lower diversity than the applicant pool: a workforce diversity "poverty trap". Our results also illustrate that if turnover rate is low, employee diversity takes a very long time to change, even in the absence of any bias. The predicted rate of change in workforce composition depends on the rate at which employees enter and leave the organization, and on three measures of inclusion: applicant diversity, appointment bias and departure bias. Quantifying these three inclusion measures is the basis of a new, practical framework to identify barriers and opportunities to increasing workforce diversity. Because we used a systems approach to investigate underlying feedback mechanisms rather than context-specific causes of low workforce diversity, our results are applicable across a wide range of settings.
ISSN:1932-6203