Beneath the Academic Verbiage: Race & Gender Biases in Customer Satisfaction Surveys

\"U966046GINP\"Guest Post: By Peter A. Bamberger, PhD, Associate Editor of the Academy of Management Journal.

Starting today we are beginning a section on this blog entitled \”Beneath the Academic Verbiage.\” In this section we will be highlighting academic articles, both past and present, that we believe have something of importance to say to proactive leaders. My colleague, Professor Bamberger, will be the contributing editor to this section.

AN EXAMINATION OF WHETHER AND HOW RACIAL AND GENDER BIASES INFLUENCE CUSTOMER SATISFACTION by: David R. Hekman, Karl Aquino, Bradley P. Owens, Terence R. Mitchell, Pauline Schilpzand, and Keith Leavitt.  To appear in a forthcoming issue of the Academy of Management Journal.

Customer satisfaction surveys have become ubiquitous.   It is nearly impossible to speak with a customer service agent without being asked to participate in a short satisfaction survey immediately upon the conclusion of the service interaction.  And recognizing the importance of customer service, an increasing number of organizations are tying employee compensation to customer ratings of agents’ service behavior.  But to what degree might these customer ratings be subject to gender or racial bias?

While noting that “U.S. society has made considerable progress in reducing overt expressions of prejudice since the Civil Rights movement during the 1960s,” Heckman and his colleagues explain how covert and unconscious forms of prejudice and bias may still, nevertheless continue to exist.  One of the explanations is that customer often have “preconceived expectations about others depending on whether the person being observed belongs to a high or low-status demographic group.”  While such biases may be held in check when raters are held accountable for their ratings, customers are rarely held accountability for their ratings of service agents.  Indeed in the vast majority of cases, ratings are provided on an anonymous basis.

To test the possibility that customer ratings are indeed subject to bias, the authors of this study designed a number of experiments to assess the degree to which: (a) the association between objective performance and customer-rated performance is weakened when the individual being rated is female or a member of a racial minority, (b) customers’ ratings of the overall firm are poorer to the degree that a highly visible employee in that environment belongs to a low (i.e., women, African-American) rather than high (i.e., male, white) status demographic group, and (c) The relationship between an organizational unit\’s objective performance and customer satisfaction is weakened as a function of the percent of unit employees belonging to low-status demographic groups (i.e., women and minorities).

Three studies were run by the authors.   In the first, customer ratings of 113 family physicians employed by a larger HMO were evaluated, with each physician rated by over 100 patients.  These customer ratings were then tested for their agreement with a number of HMO-based, objective physician performance metrics such as productivity number of customer problems visit) y, accessibility (emails sent to and received from patients), and quality (prescription writing is on par with nature of diseases treated). The study found that while that objectively-measured behaviors should have been positively related to customer satisfaction for all physicians, they were only positively related to customer satisfaction for those  physicians who were white or male.  Moreover, it found that  one type of customer-centered behavior was significantly negatively related to customer satisfaction for women and nonwhite physicians.

In the second study, participants were shown a video of actors in a book store, each of whom performed a pre-scripted interaction, with the only variable manipulated being the demographic characteristics of the agent carrying out the behavior. Using the pre-scripted interaction reduced the variability in agent behavior, thus increasing the likelihood that it was the bias causing any variance in ratings rather than differences in the ratee behavior itself that underlie any difference in ratings.   This study found that while raters were significantly less satisfied with women employees than their equally performing male counterparts, there was no evidence of such bias in customer satisfaction judgments of the nonwhite employee.  Still raters generalized about the quality of the store based on the presence of minority and women employees. More specifically, participants gave lower customer satisfaction judgments of the store environment when an employee in that environment belonged to a low (i.e., female, African-American) rather than high (i.e., male, white) status demographic group.

Finally, in the third study, the authors examined the degree to which the demographic composition (percent non-white and percent non-female) of 66 facilities of a chain of golf country club company influenced the link between objective facility performance metrics such as facility quality and productivity, and facility-specific customer satisfaction ratings.  Here too, the authors found the relationship between objective and customer-perceived ratings to be influenced by demographic factors – this time at the organizational level.  More specifically, they found a stronger positive relationship between objective performance and customer satisfaction for facilities that have a low percentage of female employees than for facilities that have a high percentage of female employees, and for facilities that have a low percentage of non-white employees relative to those having a higher percentage.

The consistency of the findings across the three different studies is remarkable and suggests that these results are most likely generalizable across contexts.  More importantly, the findings provide an important insight into the persistence of demographic inequalities in the American workplace.  More specifically they suggest that any wage or career disadvantage that minorities and women have relative to white males may stem from the rational choice on the part of managers relying upon the (biased) performance assessments of key stakeholders, namely their customers.   Accordingly, the findings also provide a clue as to why managers have often been reluctant to pursue diversity despite the known performance advantages of having a diverse workforce.   Although the study does not directly examine how the externalities of such “contextual spillover” may be avoided when assessing customer satisfaction, they do provide a number of practical suggestions such as: (1) making customer ratings less anonymous; (2) changing the standards customer use to make their ratings so they emphasize behavior rather than subjective judgments; and (3) introducing customer de-biasing education or training in the evaluation process.

By: Professor Peter A. Bamberger, PhD
Associate Editor, Academy of Management Journal
Recanati Graduate School of Business
Tel Aviv University

Let's keep in touch
Recent Posts
What They Say