In some professions, women have become well represented, yet gender bias persists-Perpetuated by those who think it is not happening

Abstract

In efforts to promote equality and combat gender bias, traditionally male-occupied professions are investing resources into hiring more women. Looking forward, if women do become well represented in a profession, does this mean equality has been achieved? Are issues of bias resolved? Two studies including a randomized double-blind experiment demonstrate that biases persist even when women become well represented (evinced in veterinary medicine). Evidence included managers evaluating an employee randomly assigned a male (versus female) name as more competent and advising a $3475.00 higher salary, equating to an 8% pay gap. Importantly, those who thought bias was not happening in their field were the key drivers of it-a "high risk" group (including men and women) that, as shown, can be readily identified/assessed. Thus, as other professions make gains in women's representation, it is vital to recognize that discrimination can persist-perpetuated by those who think it is not happening.

Fig. 1. Conceptual model illustrating that in professions once male-dominated, now with a strong representation of women, gender-biased evaluations of an employee’s competence will persist.

These biased evaluations will favor a male employee over an otherwise identical female employee and be evident among those in the profession who believe discrimination against women in their profession is no longer an issue. Biased evaluations will subsequently translate into biased treatment of the employee.

Fig. 2. Evaluations of employee competence (general measure) by the purported gender of the target employee and managers’ beliefs about whether women in their profession still face discrimination.

Scale is 1 to 7 (n = 236); higher values indicate higher competence evaluations. Among managers who believe discrimination against women is no longer an issue, the male employee was evaluated as more competent than the otherwise identical female employee. Analyses probed the interaction (by managers’ beliefs) at ±1 SD. These values correspond to a general endorsement/rejection of these beliefs. For ease of interpretation and because the values represent categorically distinct beliefs, they are presented as bars (estimated means at ±1 SD with covariates at their sample means). The differences in means correspond to the following values: “Holding the Belief that Women in the Profession Do Still Face Discrimination,” θ_X➔Y = −0.13, SE = 0.17, P = 0.44 [−0.46, 0.20]; “Holding the Belief that Women in the Profession Do NOT Face Discrimination,” θ_X➔Y = 0.48, SE = 0.17, P = 0.004 [0.15, 0.80]. For an analogous depiction with the above confidence intervals (around the conditional effect of target gender, θ_X➔Y | M), see fig. S2.

Fig. 3. Managers’ advised salary for the target employee, by the purported gender of the employee and managers’ beliefs about whether women in their profession still face discrimination (n = 229).

To account for individual differences in base salary rates, managers reported the typical salary in their practice for employees with similar experience as the target, and this was subtracted from the advised salary; thus, any differences in base salary rates (can be substantial across different regions of the country) were accounted for by analyzing respondent-specific deviations in advised salary. Y-axis values therefore represent deviations in advised salary (from individually adjusted base salary). A value of £0 indicates that managers advised paying the target employee the same as others in their practice with comparable experience. Among managers who believe discrimination against women is no longer an issue, they advised that the male employee receive a higher salary than the otherwise identical female employee. Analyses probed the interaction (by managers’ beliefs) at ±1 SD. These values correspond to a general endorsement/rejection of these beliefs. For ease of interpretation and because the values represent categorically distinct beliefs, they are presented as bars (estimated means at ±1 SD with covariates at their sample means). The differences in means correspond to the following values: “Holding the Belief that Women in the Profession Do Still Face Discrimination,” θ_X➔Y = −£303.07, SE = £832.00, P = 0.72 [−£1942.79, £1336.65]; “Holding the Belief that Women in the Profession Do NOT Face Discrimination,” θ_X➔Y = £2564.23, SE = £820.71, P = 0.002 [£946.78, £4181.69]. For an analogous depiction with the above confidence intervals (around the conditional effect of target gender, θ_X➔Y | M), see fig. S3.

Fig. 4. Managers’ differential treatment of the target employee, as a function of the employee’s purported gender and managers’ (biased) evaluations of the employee’s competence (rooted in their belief that women in the profession no longer face discrimination) (n = 222).

Managers’ competence evaluations were key to predicting how they would treat the employee {e.g., willingness to let her/him take on more supervisory responsibilities and be more involved in managing the business/financial side of the practice (if s/he was in their practice); B = 0.77 [0.56, 0.98], SE = 0.11, P < 0.001}. However, these competence evaluations were themselves systematically biased among those who thought gender bias was no longer an issue (condition*bias-belief, B = 0.22 [0.11, 0.33], SE = 0.06, P < 0.001), which translated into differential, discriminatory treatment. In other words, there was a significant indirect effect of target gender on treatment but only among those who believed that gender bias was no longer an issue: indirect effect = 0.36 [0.16, 0.62]. Among managers who rejected this belief, the employee’s gender had no bearing on how s/he would be treated (indirect effect = −0.17 [−0.38, 0.01]).