The workforce crisis part 5 – The design of the 'Mark' and 'Elizabeth' study

Does the geography of respondents in any way account for, or have a bearing on, discrepancies in advised salaries for Mark and Elizabeth?

In short, no. Before conducting the study we acknowledged that average salaries for vets may vary by geographic region. This is why we: (a) asked managers and employers to indicate what salary they would advise for ‘Mark’/’Elizabeth’ if s/he was employed in their own practice, and (b) later in the study, asked them to indicate what the typical salary is for vets in their own practice who are relatively new to the profession (graduated 1-2 years ago). This mirrors the experience level described for ‘Mark’/’Elizabeth’.

In our analyses, we then examined the size of the difference between each respondent’s advised salary for ‘Mark’/’Elizabeth’ and the typical salary for a vet of comparable experience in their own practice (and then whether across respondents this difference was consistently larger or smaller for Mark vs. Elizabeth). All to say, our analytical approach accounted for how respondent-specific differences in geographic region/location might influence base salary rates and advised salaries for ‘Mark’/’Elizabeth.’ By accounting for this, our results more precisely tap into the true level of (monetary) gender bias.

How were study respondents chosen to evaluate ‘Mark’ vs. ‘Elizabeth’?

Were they randomised, or were or they assigned to ‘Mark’ or ‘Elizabeth’ based on certain demographics?

Each respondent was randomly allocated to evaluate ‘Mark’ or ‘Elizabeth.’ Random assignment is a critical methodological element to experimental studies if one wants to test whether their manipulation (in this case, the gender of the vet) is truly what drives the pattern of results found.

At the same time, we recognize that certain demographic characteristics of respondents, such as age, gender, years of managerial experience, might influence how they evaluate vets. This is why we statistically account for these demographic characteristics in our analyses. In essence this means that our results empirically take into account any influence these demographics features might have. Thus, our analyses explain how employers/managers differentially evaluate ‘Mark’ and ‘Elizabeth’ over and above any influence that their age, gender, years of managerial experience, etc. might have on these evaluations.

What was the gender split of the respondents asked to evaluate ‘Mark’ vs. ‘Elizabeth’?

Given that respondents were randomly assigned to evaluate ‘Mark’ or ‘Elizabeth,’ we would expect the gender composition of the ‘Mark’ group to be roughly the same as those assigned to evaluate ‘Elizabeth’ and to roughly match the gender composition of the full sample of respondents. That is exactly what occurred. The full sample of respondents was 54% female; respondents who evaluated ‘Mark’ and ‘Elizabeth’ were 55% and 53% female, respectively.

Why the names ‘Mark’ and ‘Elizabeth’?

We selected names that: (a) have been used in previous studies (e.g., Hoyt & Burnette, 2013), and (b) are highly common for people born in the UK around the time that ‘Mark’ and ‘Elizabeth’ would have likely been born (mid- to late-1980’s). Both of these names are in the top 25 baby names for the mid 1980’s, according to the UK Office for National Statistics.

This approach, of selecting common names, is also consistent with other relevant research (e.g., Moss-Racusin et al., 2012). The idea is that using common names helps keep the names of the vets being evaluated fairly mundane per se.

You could imagine that using more unique names could evoke a more specific/unique set of inferences about the vet’s background, and this could in turn subtly influence respondents’ perceptions and evaluations of this vet. By selecting common names we can minimize the possibility that there is a specific and consistent set of attitudes (across the majority of respondents) or connotations attached to the name of the vet they were evaluating.

Also keep in mind that the results of this study largely coincide with: (a) evidence from a variety of other studies, using a variety of different names (for an overview, see, e.g., Koch et al., 2015), and (b) evidence of women’s own, real-world experiences with discrimination in the vet profession – and I very much doubt it’s only women in the profession named Elizabeth who are experiencing such discrimination. Altogether, I think this makes it harder to ‘explain away’ the results of this study based on the particular names used.