Occupational segregation by gender in veterinary specialties: who we’re choosing, or who’s choosing us

Abstract

Objective:

Gender demographics vary across specialties including surgery, internal medicine, cardiology, neurology, and oncology. Our objective was to determine whether residency selection or the decision to apply for training drives these differences.

Study Design:

Retrospective cohort study

Sample:

Matched and unmatched residents lists from Veterinary Internship and Residency Matching Program (VIRMP) from 2011–2020. Comparative Data Reports from the American Association of Veterinary Medical Colleges from 2010–2019.

Methods:

Names for matched and unmatched residents with addresses in the United States or Canada were coded for gender for seven programs: large and small animal surgery, large and small animal medicine, cardiology, neurology, and oncology. Match rate by gender was compared using chi-square tests. Gender demographics of applicants were compared to demographics of graduates using tests of two-proportions.

Results:

No differences were observed between genders for the likelihood of successfully matching into each residency program evaluated except in large animal internal medicine. Women (44.2%) were slightly more likely to match, overall, than men (39.0%, P=.003). The proportions of women applying for residencies overall (70.7%), in large and small animal surgery (66.1%, 62.2%), cardiology (70.2%), and neurology (70.7%) were lower than the proportion of female graduates (79%; P’s <.001).

Conclusion:

No evidence for gender bias was detected in the VIRMP resident selection process. Female veterinary graduates seemed less likely to apply for residencies than their male counterparts.

Impact:

Occupational segregation seems to stem from the decision to apply for residency. Interventions aimed at altering gender demographics in specialized medicine should target potential applicants.

INTRODUCTION

Occupational segregation is a term that describes the distribution of workers across occupations, or within specialized segments of an occupation, based on demographics. Most commonly, this refers to the over- or underrepresentation of men or women among different kinds of work. Veterinary medicine has experienced a higher rate of entry of women since the late 1980’s, when the American Association of Veterinary Medical Colleges (AAVMC) first reported more than 50% women matriculating.¹ That proportion has grown to over 80%² and has translated into a feminization of the profession itself, as women comprised 63% of all veterinary professionals in the United States in 2019.³ However, proportional dispersion of this shifting demographic is not observed in all areas of the profession. In 2020, the American College of Veterinary Surgeons (ACVS) reported that 40.5% of their diplomates were women,⁴ a 3.5% change from a 2012 census when 37% of registered diplomates were women.⁵ In 2017, the American College of Veterinary Internal Medicine (ACVIM) membership varied by specialty as follows: overall – 62.5% women, 37.5% men; large animal internal medicine (LAIM) – 64% women, 36% men; small animal internal medicine (SAIM) – 64% women, 36% men; cardiology – 54% women, 46% men; neurology – 55% women, 45% men; oncology – 69% women, 31% men.⁶ These figures suggest that all areas of specialty medicine do not experience changes to its gender demographics at a rate equivalent to the rest of the veterinary profession.

The gender-based segmentation of a professional sphere has been shown to be influenced by stereotypes associated with a particular career, or assumptions regarding a person’s suitability for the job in terms of talent or fitness, according to a person’s gender.⁷ The increasing entry of women into the workforce in the US over the past half century has not been equally distributed across professions.⁸ For example, in 2017 women accounted for 47% of the total labor force, yet represented 90% of registered nurses and 79% of elementary and middle school teachers.⁹ In contrast, women were substantially underrepresented as chief executives (28%), construction workers (9%) and software developers (19%).⁹ Occupational segregation has also been implicated as a major contributor to the gender wage gap, as an increasing share of women appears to exert a devaluing effect on the profession, driving incomes down as employers ascribe a lower value for work done by a feminized workforce.^10,11 Gender segregation in human medicine occurs across specialty fields and is strongly correlated with the average incomes of those specialties.¹² In veterinary specialty medicine large income gaps are persistent among male and female specialists.^6,13 Despite these findings, the main drivers for these demographic drifts remain unclear; whether discrimination and bias to entry mediate segregation, whether labor force participants self-select along gendered lines, or whether stereotypes or structural barriers are major effectors. The ultimate goal is to create a motivated, satisfied, and accessible profession; identifying structures that may serve as barriers or to dissuade entry for certain groups are critical steps towards reaching that goal.

The gatekeeper to most individuals seeking specialty training with the ACVS and ACVIM is the Veterinary Internship and Residency Matching Program (VIRMP). Annually, many more will apply than will successfully match into a residency training program in their chosen field. No public data currently exists to chart demographic trends among diplomates of the ACVS and ACVIM to more specifically define gender shifts within these colleges, yet a longitudinal assessment of the proportion of men and women successfully matching into residency programs would reveal the extent to which these populations are changing. Similarly, the bottleneck effect of the matching process itself represents an important area of consideration for investigating how occupational segregation occurs in specialty medicine.

Our aim in this study was to evaluate trends, and possible causes, for occupational segregation in seven specialty programs: ACVS-Large Animal (LASx), ACVS-Small Animal (SASx), ACVIM-Large Animal (LAIM), ACVIM-Small Animal (SAIM), ACVIM-Cardiology (Cardio), ACVIM-Neurology (Neuro), and ACVIM-Oncology (Onc). The first two objectives of this study were to collect data from 2011–2020 for matched and unmatched residents from the US and Canada to determine the gender demographics of successfully matched residents over a ten-year span, and to determine if there was evidence for a gender-based selection bias by comparing the demographics of the applicant pool to the matched pool. Our third objective involved collecting demographic data on graduates from US veterinary medical colleges from 2010–2019, to ascertain whether there was proportional representation of men and women among graduating populations and applicant populations to each specialty college.

MATERIALS AND METHODS

Data Collection

Methods pertaining to use of VIRMP match lists and AAVMC Comparative Data Reports were discussed with the Institutional Review Board (IRB) at the University of Wisconsin-Madison to ensure ethical use of data. Given the free availability of the data being evaluated, the IRB deemed oversight of methodology or security in handling data unwarranted, and no formal review was performed.

Matched and unmatched residents lists from the VIRMP were collected and evaluated retrospectively for ten years from 2011–2020. Only individuals with addresses in the US or Canada were included during data acquisition in order to minimize confounding factors such as VISA restrictions, education at an AVMA accredited institution, ability to obtain veterinary licensure in the US, and maintaining homogeneity with the comparison population of graduates from US Veterinary Medical Colleges. Individuals matching, or not matching, into each of the following seven specialty programs were included: SASx, LASx, SAIM, LAIM, Cardio, Neuro, and Onc. For programs with ambiguous nomenclature (i.e. “food animal medicine and surgery”), the program and/or the individual was researched to determine appropriate classification.

First names of matched and unmatched residents were used to perform manual coding of gender (JG, AP, ES). Names used commonly as female names in the United States (eg. Sophie, Heather, Jaqueline) were assigned “female” while names commonly as male names (eg. Matthew, Christopher, Kyle) were assigned “male”. Post-hoc analysis to confirm the accuracy of gender assignment for commonly used names was performed by randomly selecting 15 female and 15 male names, and subjecting them to the largest online name database for gender assignment.¹⁴ Individuals with first names of ambiguous distinction (e.g., Casey, Sam), or names that are not commonly used in the US, were subjected to an internet-based search using their full name, veterinary associated terms, and the location of their residency-matched institution or application address. Every effort was made to ensure that the results of the search matched the VIRMP candidate based on the information above. Those individuals were then identified based on phenotype from photographs, or through the use of gender pronouns referring specifically to the applicant/matched resident in question. The results of this study therefore conform to gender as assigned by name or by phenotype, and not gender identity, or an identity that is non-binary. Previous survey analysis of ACVS and ACVIM Diplomates suggests that a non-significant number of individuals identify outside of a traditional male/female paradigm (0/712 ACVS Diplomates,¹³ 1/776 ACVIM Diplomates⁶).

Annual Comparative Data Reports were obtained from the AAVMC for the ten academic years from 2009–2010 to 2018–2019. Data extracted from each report included the number of total graduates, and the number of male and female graduates, from US veterinary medical colleges. These data were then aligned with a one-year lag behind data compiled for VIRMP applicants (e.g., 2013 graduates were compared against 2014 VIRMP applicants and matched residents) in order to account for the time spent in an internship. The gender demographics of individuals applying to each of the seven specialty areas were compared with demographics of graduates from US veterinary medical colleges.

Data Analysis

If an applicant’s gender could not be identified using the methods as noted above, they were considered missing at random and were removed from the analysis. Basic descriptive statistics of applicant gender, specialty program, and match outcome were summarized by both number (N) and percentage. Match rate by gender was calculated by dividing the number of matched residents by the number of applicants within gender. Comparisons of percent match by gender were conducted using chi-square tests of association and were summarized by odds ratio (OR) and 95% confidence intervals (CI). Comparison of gender distributions between those that applied and those that graduated were conducted using tests of two independent proportions. All analyses were run in R for statistical computing version 3.5¹⁵ and at a 5% significance level.

RESULTS

Demographic trends of matched and unmatched residents

Between 2011 and 2020, 1651 individuals with addresses in the US and Canada matched into residency programs through the VIRMP; 1209 (73%) of these were coded as women and 442 (27%) were coded as men. 1524 (69%) of unmatched applicants from the US and Canada were coded as women and 692 (31%) of unmatched applicants were coded as men (2216, total). Five names could not be identified as either gender; all of these were among unmatched residents. Three were unmatched in SASx, one in SAIM, and one in Onc. When combined with the number of matched residents, an overall applicant pool of 2733 (71%) women and 1134 (29%) men (3867, total) was generated.

Over the ten years evaluated, the raw number of women matching into all residency programs was higher than the number of men matching. This trend held true overall and at every year observed, with the exception of 2014 when 53% of those who matched into LASx training programs were men (10 men vs 9 women). The overall gender demographics of matched residents during the 2011–2020 time period for each specialty area were: SASx – 36.5% men, 63.5% women; LASx – 33% men, 67% women; SAIM – 21% men, 79% women; LAIM – 16.5% men, 83.5% women; Cardiology – 30% men, 70% women; Neurology – 27% men, 73% women; Oncology – 20% men, 80% women. A complete listing of numbers, and percentages, of matched residents by gender, specialty college, and year are provided in Supplementary Item 1.

Matched resident data

Over the ten years evaluated, there were no significant differences in the match rate by gender into residencies for the following programs: SASx, LASx, SAIM, Cardiology, Neurology, and Oncology (Figure 1, Table 1). Women were slightly more likely than men to match into LAIM residencies overall (71.2% vs 56.2%; P = .049, OR 1.92 [1.00, 3.69]; Figure 1), but this difference was not significant at any individual time point (year). Women were more likely than men to match into SASx training programs in 2020 (55% vs. 25%; P = .005, OR 3.67 [1.48, 9.11]) and into SAIM training programs in 2015 (62.1% vs 30.8%; P = .048, OR 3.68 [1.01, 13.4]).

Figure 1.

Percentage of applicants successfully matching by gender into each specialty program from 2011– 2020. The total number of men and women successfully matching into each program over that time period is listed along the x-axis (Total N). P-values represent comparisons of percent match by gender. *P < 0.05 is significant.

Table 1.

Comparison of the match rate by gender for seven specialty programs over the 2011–2020 time period.

	Gender	N	% (95% CI)	OR (95% CI)	P-Value
SASx	Men Women	437 719	35.5 (31.1, 40.1) 37.4 (33.9, 41.0)	reference 1.09 [0.85–1.39]	0.506
LASx	Men Women	195 381	32.8 (26.6, 39.7) 34.9 (30.3, 39.8)	reference 1.10 [0.76–1.58]	0.617
SAIM	Men Women	159 585	53.5 (45.7, 61.1) 55.4 (51.3, 59.4)	reference 1.08 [0.76–1.54]	0.665
LAIM	Men Women	48 191	56.3 (42.1, 69.5) 71.2 (64.4, 77.2)	reference 1.92 [1.00–3.69]	0.049*
Cardiology	Men Women	110 259	27.3 (19.8, 36.3) 27.4 (22.3, 33.2)	reference 1.01 [0.61–1.66]	0.978
Neurology	Men Women	107 258	38.3 (29.6, 47.8) 43.8 (37.9, 49.9)	reference 1.25 [0.79–1.99]	0.335
Oncology	Men Women	78 340	51.3 (40.3, 62.1) 47.9 (42.7, 53.3)	reference 0.87 [0.53–1.43]	0.595
Total	Men Women	1134 2733	39.0 (36.2, 41.8) 44.2 (42.4, 46.1)	reference 1.24 [1.08–1.43]	0.003*

N: number of applicants from all years evaluated

%: percent of gender successfully matching into program

^*P <0.05 is significant

Over the timespan of our analysis, and across all specialties evaluated, women were slightly more likely, overall, to match into any ACVS or ACVIM accredited residency program compared with men (44.2% vs 39%; P = .003, OR 1.24 [1.08, 1.43]; Table 1). This difference was significant only in 2020, when women were more likely to match into ACVS or ACVIM residencies overall than men (52.6% vs 35.7%; P = .003, OR 1.99 [1.27, 3.14]; Table 2). A complete listing of match rates by specialty college and by year is provided in Supplementary Item 1.

Table 2.

Comparisons of the match rate by gender into all seven specialty programs evaluated, by year.

	Gender	N	% (95% CI)	OR (95% CI)	P-Value
2011	Men Women	102 254	30.4 (22.3, 40.0) 36.2 (30.5, 42.3)	reference 1.30 [0.79–2.13]	0.296
2012	Men Women	118 248	33.1 (25.2, 42) 43.1 (37.1, 49.4)	reference 1.54 [0.97–2.43]	0.066
2013	Men Women	117 267	37.6 (29.3, 46.7) 40.4 (34.7, 46.4)	reference 1.13 [0.72–1.76]	0.600
2014	Men Women	101 259	41.6 (32.4, 51.4) 43.2 (37.3, 49.3)	reference 1.07 [0.67–1.71]	0.775
2015	Men Women	108 252	38.9 (30.2, 48.4) 47.2 (41.1, 53.4)	reference 1.41 [0.89–2.23]	0.146
2016	Men Women	113 283	47.8 (38.8, 57.0) 44.5 (38.8, 50.4)	reference 0.88 [0.57–1.36]	0.556
2017	Men Women	113 311	35.4 (27.1, 44.6) 44.4 (38.9, 49.9)	reference 1.46 [0.93–2.27]	0.099
2018	Men Women	117 308	41.0 (32.5, 50.1) 41.9 (36.5, 47.5)	reference 1.04 [0.67–1.60]	0.873
2019	Men Women	133 277	46.6 (38.3, 55.1) 48.4 (42.5, 54.3)	reference 1.07 [0.71–1.62]	0.739
2020	Men Women	112 274	35.7 (27.4, 45.0) 52.6 (46.6, 58.6)	reference 1.99 [1.27–3.14]	0.003*

N: number of applicants from all programs evaluated

%: percent of gender successfully matching into all programs

^*P <0.05 is significant

Applicant demographics versus graduate demographics

From 2010–2019, 6249 (20.5%) of graduates from US veterinary medical colleges were men, and 24,179 (79.5%) were women (30,426, total). There was no overall difference between the male-to-female ratio of graduates and the male-to-female ratio of applicants for SAIM, LAIM, and Oncology (Figure 2). The male-to-female ratio of graduates was significantly different than the male-to-female ratio of applicants for SASx (P < .001), LASx (P < .001), Cardiology (P < .001), Neurology (P < .001), and overall for all applicants to all programs evaluated (P < .001; Figure 2). In all cases, men were proportionally overrepresented and women were proportionally underrepresented within the residency application population. These disparities were significantly different at various time points for each of the following specialties; SASx – all years (2011, P = .003; 2012–2020, P < .001); LASx – 2012 (P = .037), 2013 (P = .0429), 2014 (P = .001), 2017 (P < .001), 2018 (P = .006), 2019 (P = .042), 2020 (P = .015); Cardiology – 2011 (P = .029), 2012 (P = .042); Neurology – 2015 (P = .003), 2016 (P = .011); SAIM – 2019 (P= .041). A complete dataset comparing all applicant and graduate demographics across specialty colleges and years is provided in Supplementary Item 1.

Figure 2.

Percentage of men and women among total residency applicants for each specialty program between 2011–2020 (vertical bars) versus the average percentage of men and women graduating from US Veterinary medical colleges between 2010–2019 (dotted lines). Total number of male and female applicants for each specialty program are listed along the x-axis (Total N). P-values represent comparisons of gender distributions between those that applied and those that graduated. *P < 0.05 is significant.

DISCUSSION

In this study we describe the gender demographics of residents entering specialty training programs with the ACVS and ACVIM between 2011 and 2020 to better understand dynamic shifts across the profession. Our data reveal a shift towards an increasingly feminized diplomate workforce across all specialties evaluated, compared with current college demographics. This increased entry of women is an opportunity to consider structural adaptations of working environments to better serve a changing labor pool. We also identified variable occupational segregation across specialties, where men are proportionally overrepresented in the field of surgery (SASx, LASx), as well as in specialties where surgical skills are a required component of training and/or practice (Cardiology, Neurology). Prior to this investigation no longitudinal quantification of this dispersion existed, and the processes shaping these differences were unclear. In other words, is the residency selection process, or the decision to apply for a specific specialty, driving the observed variations? While we found no difference in match rate by gender into training programs, comparison of graduation demographics to applicant statistics suggests this distribution is more likely driven by how individuals segregate and make choices to apply for residency, rather than a primary selection bias that could occur during the VIRMP match process. An ideal, or targeted, demographic distribution for specialty groups does not exist, as the workforce needs to balance representation of both the professional population it is drawn from, as well as the wider population that it serves. The observation that the majority (female) graduate population is less likely to pursue some fields, however, suggests a contraction of the talent pool, and the existence of either barriers or factors that may deter interest. In an effort to create the most attractive and sustainable careers, these findings highlight areas of interest and consideration.

Specialty fields are becoming increasingly female, in alignment with the overarching trend of the veterinary profession. Women comprised between 63.5 and 83.5% of individuals who matched into residency programs over the past ten years. While large and small animal surgery appear poised for a demographic shift to a female majority, other specialty areas have already tipped that scale to varying degrees but are experiencing continued growth of women in practice. The workforce is changing and will continue to look different than it does now; while specialization can be an attractive option for reasons outlined previously,^6,13,16,17 it poses significant points to consider for private practices and academic settings, alike. The needs and experiences of a female-predominant workforce are likely to be different on professional and personal levels, including work hours, career decisions, income and advancement gaps, and requirements for pregnancy and family planning.^6,13,16–18 The degree to which a working environment will need to adapt may change as the proportion of women in the workforce continues to grow.

Occupational feminization is associated with lower incomes and a change in the external perceptions of a profession, as an effect of the stereotypes assigned to the predominant gender and/or the perception that women may be less competent than men.^11,12,19 The vocation may be stigmatized as “women’s work” and consequently devalued in compensations.^20,21 As a result, men may be deterred from entering the profession. Lincoln exhibited a general stagnation in wages relative to four other comparison occupations (physicians, lawyers, optometrists, pharmacists), as well as a temporal association between that relative decline in veterinary earnings and fewer applications from men to veterinary school.²² The three small animal specialist groups with the highest proportions of men currently reported are SASx, cardiology, and neurology; we demonstrated these three specialties receiving significantly higher proportions of men applying for residency programs as compared with graduating from veterinary school. One possible explanation for the increased interest among men is the comparatively higher salaries achieved by men in those areas of practice;^6,13 median incomes reported for male small animal surgeons, cardiologists, or neurologists in private practice were the highest reported for any specialist of any gender. While the services provided and skills required across specialties both drive the generation of income and interest of young veterinarians, it is impossible to rule out the influence of feminization and devaluation on incomes and gender shifts within specialty medicine. Pelley and colleagues provided evidence for a strong correlation between female share of a medical specialty workforce and average salary, such that the female share of a specialty explained 64% of the difference in salaries observed.¹² A significant income gap between men and women veterinary specialists has been well documented; these disparities are observed to be somewhat larger in the ACVIM, already a predominantly female group, compared with the male-dominated ACVS.^6,13 The economic effects of demographic shifts are difficult to project, yet these issues are of interest to specialists as well as managers and practice owners observing practice revenue. Continued accumulation of longitudinal data on those who enter, as well as wages, will help discern how these trends intersect.

We were able to determine that neither gender was more or less likely to successfully match than the other by comparing the number of men and women applying to residency programs to those successfully matching into residency programs,. The exception was among LAIM applications, where women were slightly more likely to match than men. This finding should not be overinterpreted, given an odds ratio of 1.92, and a sample size that was heavily skewed towards women (191 women, 48 men). Similarly, while women were slightly more likely to match overall than men (44.2% women, 39% men, OR 1.24), the comparison provides weak evidence for relative risk and should be considered cautiously. These findings are encouraging in terms of concern about bias in application selection; women are no less, and possibly more, likely than men, to be selected through the VIRMP match, and the match process itself is unlikely to be affected – at least on a large scale – by gender bias. This is especially reassuring in light of relatively smaller populations of male applicants and anecdotal reports of a desire to maintain a more gender-balanced work environment. One example of a profession wide effort to maintain gender diversity includes matriculant selection at US veterinary medical colleges, where a greater proportion of men consistently enroll nationally as first year veterinary students compared with the applicant pool.^2,23,24 Several studies have shown that evaluations and applications are subject to implicit bias from those who write evaluations and recommendations and from those who review them;^25–28 in either instance, this may have a direct effect on final selection or award. Outlier cases of selection bias could possibly be identified on an institutional or programmatic level, but it is not evident at scale - suggesting that bias does not have an overriding influence on either the match itself or most of those participating in candidate evaluation and selection. Still, efforts towards obtaining and maintaining bias literacy and accountability are important to prevent its manifestation in other areas of training and practice.

We were able to identify trends in how men and women were more or less likely to apply for various residency programs by more broadly considering the residency applicant pool as a conditional subset of those graduating from veterinary medical colleges. Through these observations we propose a likely cause for occupational segregation; men apply, and are selected for, surgery, cardiology, and neurology residencies, and residencies overall, in higher proportions than they graduate in, and the converse is true for women. In other words, individuals self-segregate along gendered lines through the decision to apply for residency training in a certain field. Therefore, any ability to influence the gender demographics of a professional space would have to target individuals prior to the application process, in a way that would motivate them and prepare them to produce a competitive application. We do not clearly understand what drives choices along variably gendered lines, or what has influenced those decisions along the way. One study of graduates from US medical schools showed that female students were more likely to enter into residency programs with a higher proportion of women residents already present,²⁹ possibly due to better support structures for women, or a preference to be surrounded with more of their own gender. We observe the same trend among men in this study, though it is not known if similar explanations would hold true. Overall, women applied for residencies in smaller proportions than they graduated from veterinary school. Women may more strongly consider the interrelationship of work and time when determining career choice as compared with men.³⁰ Female veterinary specialists both experience and perceive a more negative interrelationship between family and career;^16,17 those experiences have caused some women physicians to leave the field of surgical practice.^31,32 Another study revealed higher rates of entry of women into residency training programs in specialties that decreased their working hours, and showed that hour requirements influenced directly influenced female doctor’s career choices more than other factors.³³ This same study provided some evidence for improved fertility rates among women doctors in positions with hour constraints. Labor economist Claudia Goldin posits that women in the US workforce tend to select positions with greater time flexibility,¹⁹ and therefore work hour reform may be an effective adaptation to increased entry of women into a profession. Surgery, in particular, is associated with higher on-call burdens than other specialties.^6,13 The simple observation of structural barriers or a demanding work schedule may deter female students and interns from pursuing higher level training at a higher rate than men.

The element of bias is possibly more impactful at shaping the applicant pool than at influencing the selection process. A competitive application requires competitive grades, ample experience, quality letters, and very often, a research record. All of these requirements are driven by the formation of strong relationships with specialists early on during one’s veterinary school training. Characteristically “male stereotypes,” or agentic traits often associated with men, and masculine images are consistently associated with what people perceive as a successful scientist than with female, or communal, traits.^7,34,35 Innate brilliance is implicitly more commonly associated with men than with women, and is also a characteristic that tends to be highly valued and sought after in the fields of science and technology.^36,37 Engaging in these patterns of thought, even unwittingly, confirms validity and ultimately creates an unequal playing field - boosting confidences among some, while diminishing it among others. These effects may manifest more strongly in certain specialties where the skillsets required may align more closely with various gender stereotypes. There are innumerable reasons that individuals may choose a particular specialty over another; it is possible that surgical skillsets, required in LASx, SASx, cardiology, and neurology, are associated with stereotypically male, or agentic, qualities such as confidence, technical skill, and ambition, whereas medicine-only fields such as SAIM, LAIM, and oncology, require stereotypically female, or communal qualities, such as humanitarianism, kindness, and helpfulness. Our results demonstrated significant gender shifts from graduate pools to applicant pools for each of these specialties. Similar associations are seen in human medical fields where men congregate in surgical specialties, and women in internal medicine and pediatrics.¹² Self-selection at the time of application is likely the predominant driving force. This process can be driven simply by interest in the field, by self-confidence, self-efficacy, or by role congruency. Career choices may also be influenced when students and interns that have benefitted from mentoring and training have also been subject to an implicit selection process along the way. Bias training and interventions may be more effective when targeting how clinicians interact with and influence students, in order to shape the future demographics of specialty colleges. Too, students themselves may conform to this role-congruency, and efforts at providing opportunities and instilling confidence can help to overturn traditional stereotypes. Finally, one must consider that myriad factors influence the decision to pursue a career path, and there are endless ways to consider the differences among specialties. Professional interests, personal happiness, family, goal structures, the desire to spend more years in training, financial needs and motivations, job availability, and geographic preferences, all contribute to choices to varying degrees, and it would be difficult to assess these factors for each individual without in depth qualitative research in this specific population of veterinary students and specialists.

Our data is not without limitation. Our analyses are based on a theoretical, but unrealistic, assumption that an applicant in any given year is applying for the first time, and that they apply after one year of internship. Each year was treated as an independent sample and, given that no other applicant qualifications were assessed, any consideration for reapplication was unwarranted. Internship data was not collected and was not considered as part of a longitudinal analysis, to more completely examine the pipeline for how men and women navigate careers following veterinary school. In addition to the complex nature of pursuing multiple internships, there are the confounding factors of specialty internships, and internships generated outside the match – which include the vast majority of equine private practice internships. Veterinary medical college graduate data was used as a baseline comparator for resident applicant data instead of internship data, as it provided the most complete picture of the population from which future residents will be derived, and to which interventions might be applied. We could not assess applicant quality (e.g., credentials, experience, letters of recommendation, rank, GPA) based on the information available to us. While possible, applicant quality was unlikely to differ significantly between two large sample populations solely on the basis of gender. As match rates were not significantly different among men and women, any difference in quality would suggest an implicit gender bias could exist and would favor the selection of whichever group was shown to be less qualified. Further, much of the data provided in the VIRMP application, outside of class rank and GPA, is open to subjective interpretation, including quality of training programs, interpretation of letters of recommendation, and value of research completed among others. Given the variability in assessing selection criteria, this approach would be problematic.³⁸ Our sample does not include programs taking residents outside of the match; therefore, our results may not accurately describe the behavior of programs outside the VIRMP, or specialties not evaluated in this study. Finally, gender phenotype was assigned by name or visual identification, and other identity characteristics – such as race or ethnicity or gender identity – could not be identified in an accurate or unbiased manner. Selection and application dynamics might possibly follow similar patterns for other identity groups, but this is unknown. This data is currently unavailable; however specialty medicine appears to be at a critical juncture in its evolution and should prioritize obtaining and considering this information, including collection of data that embraces race, ethnicity, and non-binary identities that have historically experienced marginalization.

Specialty medicine is feminizing and is doing so at varying rates. While bias does not appear to play a significant role in resident selection, demographic shifts appear to be driven by application pools, demonstrating that interventions should target how and when we, as specialists, interact with students and interns if we intend to influence career paths. These findings are valuable for practitioners, managers, and owners, to determine in which ways practice settings can adjust their structure and their climate to meet the changing needs of their workforce. These findings are also critical for educators, administrators, those participating in residency recruitment and training, and ACVS and ACVIM diplomates as a whole, to consider how and who they inspire and encourage as young veterinarians to step into the shoes of the future of specialty veterinary practice.