Secondary Logo

Journal Logo

AAPA Members can view Full text articles for FREE. Not a Member? Join today!
Original Research

Reexamining the persisting wage gap between male and female PAs

McCall, Timothy C. PhD; Smith, Noël E. MA

Author Information
Journal of the American Academy of Physician Assistants: November 2020 - Volume 33 - Issue 11 - p 38-42
doi: 10.1097/01.JAA.0000718284.35516.87
  • Free

Abstract

A wage gap between men and women in the United States exists, and in recent years, more attention has been given to the issue in healthcare, in both the physician and the physician assistant (PA) workforce.1-6 Coombs and Valentin examined salary differentials among PA educators; Coplan and colleagues examined a wider PA workforce and controlled for specialty, experience, number of hours worked and patients seen, and number of on-call hours.4,5 They found that male PAs consistently received higher compensation than female PAs, across a variety of specialties and after controlling for the aforementioned variables.5 Smith and colleagues controlled for the same variables as Coplan and colleagues and refined the model to include postgraduate training experience, whether PAs received a bonus, and other factors.6 Once again, there was a significant measurable difference in the overall compensation for male and female PAs.6

In popular media, this wage gap often is reported in raw dollar differences, such as “women are compensated 80% for every dollar men are compensated.”1 Although true, many of these analyses do not statistically control for real occupational differences between men and women that account for a portion of the difference. When these factors are controlled, the wage gap generally is smaller.

This study accounted for various compensation models in healthcare, including productivity-based models, hourly models, and several other demographics not considered in past models. This broader dataset makes the study a more robust model for PA wage analyses.

METHODS

Survey and respondents

Participants were selected from AAPA's database of PAs (consisting of members and nonmembers) and recruited through social media, AAPA emails, and via AAPA.org to take the 2019 AAPA Salary Survey in February 2019. The survey was open to all nonretired, US-based PAs and included a battery of personal and workplace demographics, compensation, and benefits for calendar year 2018. A subset of data from this survey was analyzed. A total of 13,088 partial or complete responses were collected from PAs. To be included in this analysis, the respondent must have completed each question relevant to this work; 8,339 respondents were included.

The study is exempt from institutional review board review in accordance with the US Department of Health and Human Service's policy for protection of human research subjects.

Data cleaning and transformations

Each respondent's base compensation was transformed into an annual wage. PAs who received a base salary or productivity pay (for example, $107,500 per year) remained unchanged. Hourly wages were annualized by multiplying the hourly wage by number of weeks worked in the year, and then multiplied by hours worked weekly. For base compensation, $450,000 was used as an upper limit, and $15,000 was used as a lower limit, resulting in less than 0.5% of wages being excluded from the analysis.

For each respondent, total compensation was calculated as the sum of base compensation and bonus ($0 if no bonus was received). Compensation data generally are positively skewed and nonnormal. Many analysts log-transform wage data to correct for skewed data, though the distribution of the data was not improved substantially. As such, and to aid in interpreting regression coefficients, wages were analyzed in their original scale of 2019 US dollars.

Statistical analysis

For this analysis, all respondents provided the following: sex, highest level of education completed, race, ethnicity, geographic region based on state, statewide cost of living, mode of compensation, amount of base compensation, whether a bonus was received and the amount, years of experience, primary specialty, primary work setting, hours worked weekly for primary employer, weeks worked annually for primary employer, patients per week, whether a PA took call in the previous year, whether a PA was in a formal leadership role in the previous year (this could include, but was not limited to, PAs who had administrative roles), and ownership in a practice.

To conclude whether statistically significant differences existed between male and female PAs on a variety of workplace factors, t-tests were used for continuous variables. Variables included in the t-test analyses included total compensation, bonus amount (if received), years of experience, hours worked weekly, weeks worked annually, and patients seen per week; Cohen's d, a measure of effect size, was computed for each comparison. Z-tests of column proportions were used for categorical variables to determine statistically significant differences between male and female PAs for factors including mode of compensation, whether a PA took call, whether a PA was in formal leadership, and whether a PA owned or shared ownership in practice.

To analyze the relationship between total compensation and sex, a multiple sequential regression was used to predict total compensation from a variety of factors: highest level of education completed, race, ethnicity, geographic region based on state, cost of living at the state level, mode of compensation, whether a bonus was received, years of experience, primary specialty, primary work setting, hours worked weekly for primary employer, weeks worked annually for primary employer, patients per week, whether a PA took call, formal leadership roles, ownership in a practice, and finally, sex. To determine whether a wage gap between male and female PAs over time may become larger or smaller over time, an interaction term between sex and years of experience also was included in the regression model.

For the multiple sequential regression, all predictors other than sex and the interaction between sex and years of experience were included in the first step of the regression. In the second step, sex and the interaction between sex and years of experience also were included in the model. This analytical procedure was used to assess two questions: (1) which factors were significantly related to total compensation while simultaneously controlling for all other factors, and (2) did sex predict total compensation above and beyond all other predictors. The final model that included sex and the interaction with years of experience is reported.

RESULTS

Respondent characteristics

The final sample for analysis was 69% female, with a mean age of 38.8 years, and included a representative sampling of primary care (22%), internal medicine subspecialties (12.1%), pediatric subspecialties (1.4%), surgical subspecialties (27.4%), emergency medicine (8.7%), no medical specialty (1.3%), and all other specialties (27%). Primary work settings represented were physician office or outpatient clinic (55.2%), hospital (35%), urgent care center (5.8%), and other (4%).

Base compensation, bonus, and total compensation by sex

Among all PAs, an analysis was conducted to determine differences by sex in base compensation, bonus, and the sum of those numbers, total compensation. Among all PAs, mean base compensation was $110,842.78; female PAs received a mean base compensation of $105,532.85 and male PAs received a mean base compensation of $122,691.66, 95 CI of the difference [$15,369.78, $18,947.83], P < .001, t(8277) = 18.80, Cohen d = 0.46. Among the full sample, which included part-time PAs, median compensation was $105,476, which we benchmarked to data from the US Bureau of Labor Statistics from 2018. That benchmark indicated median annualized full-time PA wages were $108,610.7

Among all PAs, 49% received a bonus; female PAs were less likely to receive a bonus than male PAs (47.2% of female PAs versus 53.1% of male PAs), P < .001, chi-square = 25.01. Among all PAs, the mean bonus was $11,430.83; female PAs received a mean bonus of $10,000.29 and male PAs received a mean bonus of $14,222.19, 95 CI of the difference [$3,027.40, $5,416.41], P < .001, t(3994) = 6.93, Cohen d = 0.23.

Among all PAs, mean total compensation was $116,209.68; female PAs received a mean total compensation of $109,986.87 and male PAs received a mean total compensation of $130,091.71, 95 CI of the difference [$18,111.85, $22,097.84], P < .001, t(8298) = 19.78, Cohen d = 0.51.

Workplace demographics by sex

Differences between male and female PAs on a variety of factors and statistically significant differences from a z-test of column proportions are noted in Table 1. In 2018, female PAs were significantly less likely to be paid based on productivity and significantly less likely to take call, to be in formal leadership, or to own or share ownership in a practice. In addition, male and female PAs significantly differed in work experience and time worked: female PAs had fewer years of experience, worked fewer hours weekly, and worked fewer weeks in the year at their primary employer. They also saw fewer patients in a typical week, compared with male PAs.

TABLE 1. - Select compensation-relevant factors by sex
Based on 8,339 respondents analyzed in the final sample. For statistical significance, ∗∗∗ = P < .001, ∗∗ = P < .01, ∗ = P < .05, for either a z-test of column proportions (for percentages) or a t-test (for means).
Variables All PAs (%) Female PAs (%) Male PAs (%) Significance level
Mode of compensation
  Base salary 75.4 75.9 74.1
  Hourly 20.5 20.2 20.9
  Productivity pay 4.2 3.9 4.9
PA took call last year 34.9 32.3 40.7 ∗∗∗
PA is in formal leadership 10.4 8.1 15.4 ∗∗∗
PA owns or shares ownership in practice 1.4 1 2.4 ∗∗∗
Variables All PAs(mean) Female PAs(mean) Male PAs(mean) Significance level
Years of experience 10.27 9.54 11.88 ∗∗∗
Hours worked weekly 44.53 43.65 46.49 ∗∗∗
Weeks worked last year 43.82 43.56 44.4 ∗∗∗
Patients per week 66.39 64.4 70.82 ∗∗∗

Sequential multiple regression analysis

A sequential multiple regression was conducted with all variables of interest entered into the model before sex. Sex and the interaction between sex and years of experience were entered into the model after controlling for all other variables. Table 2 includes the unstandardized regression coefficients, standard errors, 95% confidence intervals for the regression coefficient, and significance levels of each predictor in the model. Factors such as highest level of education completed, race, and ethnicity were significant predictors of total compensation in simple linear regression models but were no longer significant in a multiple regression model that included other variables. The regression model significantly predicted total compensation, P < .001, F(34, 8,162) = 126.21. The final model adjusted R2 = 0.342, P < .001; 34.2% of the variance in total compensation was explained by the factors in the regression model, and the change in R2 was significant when sex was added to the model (P < .001), indicating that sex predicted total compensation even after controlling for all other predictors.

Table 2. - Regression analysis of PA total compensation (full- and part-time PAs)
Figures are in dollars. Thirty-four percent of the variance (adjusted R2) in total compensation was accounted for by the model. Final model R2 = 0.34, with R2 change for sex significant at P < .001. Predicted total compensation based on regression model and analysis of covariance was $113,403.59 for female PAs and $122,413.17 for male PAs, a difference of $9,010, or 92.6% women/men. For statistical significance, ∗∗∗ = P < .001, ∗∗ = P < .01, ∗ = P < .05.
Variables Coefficient (B) 95% CI Standard error Significance level
Highest level of education completed (reference: master’s degree)
Associate’s -4,114.62 -11,637.28, 3,298.21 3,835.09
Bachelor’s 1,746.09 -782.29, 4,274.46 1,289.82
Doctoral 2,888.81 -2,526.84, 8,304.46 2,762.73
Race (reference: White)
Black 137.63 -4,584.50, 4,859.76 2,408.94
American Indian or Alaskan Native 5,699.76 -6,713.44, 18,112.95 6,332.44
Asian 1,417.10 -1,924.55, 4,758.75 1,704.70
Native Hawaiian or other Pacific Islander -8,808.63 -27,389.21, 9,771.95 9,478.66
Other 672.68 -5,841.16, 7,186.52 3,322.96
Two or more races -872.85 -6,041.26, 4,295.56 2,636.60
Ethnicity: Hispanic -656.68 -4,171.87, 2,858.52 1,793.23
Geographic region (reference: Midwest)
Northeast -8,618.21 -11,292.85, -5,943.57 1,364.44 ∗∗∗
Southern -1,114.49 -3,051.34, 822.36 988.06
Western 3,400.30 982.34, 5,818.27 1,233.49
Mode of compensation (reference: base salary)
Annualized hourly wage -7,595.51 -9,616.12, -5,574.90 1,030.79 ∗∗∗
Productivity pay 61,293.18 57,578.86, 65,007.49 1,894.81 ∗∗∗
Additional compensation and cost of living
Bonus received 6,941.55 5,610.69, 8,272.42 678.92 ∗∗∗
2018 cost-of-living index (Council for Community and Economic Research) 339.46 281.85, 397.06 29.39 ∗∗∗
Work experience
Years of experience 827.86 694.54, 961.38 68.06 ∗∗∗
Hours worked weekly (primary employer) 467.74 422.64, 512.84 23.01 ∗∗∗
Weeks worked last year (primary employer) 591.15 515.69, 666.60 38.49 ∗∗∗
Patients per week (primary employer) 138.28 117.50, 159.05 10.60 ∗∗∗
PA took call 3,318.21 1,715.23, 4.921.19 817.74 ∗∗∗
PA is in a formal leadership role 11,693.66 9,288.18, 14,099.13 1,227.12 ∗∗∗
PA owns or shares ownership in practice 20,034.08 13,895.08, 26,173.08 3,131.74 ∗∗∗
Primary major specialty area (reference: primary care)
Internal medicine 7,726.85 5,047.72, 10,405.99 1,366.73 ∗∗∗
Pediatric subspecialties 6,325.14 163.18, 12,487.11 3,143.45
Surgical subspecialties 12,663.20 10,447.73, 14,878.67 1,130.20 ∗∗∗
Emergency medicine 19,516.41 16,035.82, 22,997.00 1,775.58 ∗∗∗
Other 9,591.06 7,387.06, 11,795.07 1,124.35 ∗∗∗
No medical specialty 6,807.87 485.40, 13,130.34 3,225.33
Primary work setting (reference: physician office or clinic)
Hospital 8,199.70 6,299.25, 10,100.14 969.49 ∗∗∗
Other -258.30 -2,949.16, 2,432.55 1,372.71
Sex
Female -9,009.58 -11,378.59, -6,640.57 1,208.52 ∗∗∗
Female x years of experience -201.90 -365.12, -38.68 83.26

Predicted total compensation based on the regression model and analysis of covariance was $113,403.59 for female PAs and $122,413.17 for male PAs, a difference of $9,009.58, P < .001, 95 CI [$6,640.57, $11,378.59]. Among PAs new to the profession, female PAs were paid almost $0.93 for every dollar male PAs were paid. Moreover, the interaction between sex and years of experience was significant; for each additional year of work experience a PA had, the wage gap between male and female PAs widened by $201.90, P = .015, 95 CI [$38.68, $365.12].

DISCUSSION

This work builds on previous research by the authors and others and gives a more complete picture of the wage gap between male and female PAs. After calendar year 2018, a wage gap between male and female PAs persisted, even after controlling for a variety of compensation-relevant factors. Although the unadjusted wage gap between male and female PAs is 15%, it shrinks to around 7.5% when accounting for factors other than sex, and this gap widens by $201 for each additional year of work experience.

This analysis demonstrates that total compensation among female PAs was less than male PAs even after statistically controlling for more than a dozen factors, including highest level of education completed, race, ethnicity, geographic region of work, statewide cost of living, mode of compensation, whether a bonus was received, years of experience, primary specialty, primary work setting, hours worked weekly, weeks worked annually, patients seen weekly, whether a PA took call, leadership roles, and ownership in a practice. All of those factors in the sequential multiple regression model were significant predictors of wage, with the exception of education, race, and ethnicity. Expanding on the work of Kang and colleagues, our research concludes that disparities exist, even when statistically accounting for real occupational differences between men and women.8

LIMITATIONS

This study is not without limitations. Although the sample analyzed was representative of the age, sex, and specialization of PAs across the United States, the potential still exists for self-reporting bias, nonresponse bias due to a low response rate of online nationwide surveys, and issues of sample representativeness. To assess the extent these factors could have played, we benchmarked our sample to data from the US Bureau of Labor Statistics (BLS) for 2018. This allowed for a comparison to a second data source. BLS data for 2018 indicated median annualized full-time PA wages were $108,610.7 This study sample's median compensation was very close to that estimate, and included some part-time PAs, with a median of $105,476, increasing our confidence that self-reporting and issues of representativeness and nonresponse were minimal.

A second limitation to this research is the inability of the authors to include unknown or unmeasured variables in the statistical model. The role that other unmeasured variables play in explaining wages between men and women cannot be discounted. It is likely that other predictive variables exist that were unmeasured.

Similarly, a third limitation is the inability of the authors to conclusively demonstrate that the remaining wage disparity between male and female PAs is due to explicit discrimination or some other factor. Rather, the authors conclude that after controlling for compensation-relevant factors available for analysis, a wage gap between male and female PAs persists.

FUTURE DIRECTIONS

Researchers should explore other unmeasured factors that may explain a portion of this difference. In addition, researchers should explore lifelong implications and cost estimates of wage gaps between male and female PAs, given our research indicates the wage gap worsens with additional work experience.

Several policy implications also have the potential to mitigate wage disparities, and some have been used by various organizations, agencies, and locales. These include not inquiring about previous wages when hiring (which could perpetuate and exacerbate disparities over time), laws requiring publication of compensation statistics for companies, statistical self-audits within organizations that use regression analyses similar to those employed in this study, pay range standardization or banding such as those in government agencies, and other ways to reduce managerial discretion in wages such as contracting with outside consultants to price wages for positions within an organization.9-12 Steps such as these do not guarantee success for organizations working to reduce their wage gaps, but there has been some evidence of their success.13,14

CONCLUSION

A wage gap between male and female PAs persists, and wage parity remains a goal for PAs. After accounting for a variety of factors that affect compensation for men and women in calendar year 2018, female PAs were paid almost $0.93 for every $1 male PAs received, and this gap widened with each additional year of work experience. Both employees and employers should consider and implement steps to lessen the wage gap, which amounts to thousands of dollars annually for female PAs.

REFERENCES

1. American Association of University Women. The simple truth about the gender pay gap: Fall 2018 edition. www.aauw.org/resources/research/simple-truth. Accessed June 23, 2020.
2. Jena AB, Olenski AR, Blumenthal DM. Sex differences in physician salary in US public medical schools. JAMA Intern Med. 2016;176(9):1294–1304.
3. Seabury S, Chandra A, Jena A. Trends in the earnings of male and female health care professionals in the United States, 1987 to 2010. JAMA Intern Med. 2013;173(18):1748–1750.
    4. Coombs J, Valentin V. Salary differences of male and female physician assistant educators. J Physician Assist Educ. 2014;25(3):9–14.
    5. Coplan B, Essary AC, Virden TB III, et al. Salary discrepancies between practicing male and female physician assistants. Womens Health Issues. 2012;22(1):e83–e89.
    6. Smith N, Cawley JF, McCall TC. Examining the gap: compensation disparities between male and female physician assistants. Womens Health Issues. 2017;27(5):607–613.
    7. US Department of Labor. Bureau of Labor Statistics. Occupational employment statistics. www.bls.gov/oes. Accessed June 23, 2020.
    8. Kang SK, Kaplan S. Working toward gender diversity and inclusion in medicine: myths and solutions. Lancet. 2019;393(10171):579–586.
    9. Bennedsen M, Simintzi E, Tsoutsoura M, Wolfenzon D. Do firms respond to gender pay gap transparency? National Bureau of Economic Research Working Paper Series No. 25435. January 2019. www.nber.org/papers/w25435. Accessed July 27, 2020.
    10. Society for Human Resource Management. Managing pay equity. www.shrm.org/resourcesandtools/tools-and-samples/toolkits/pages/managingpayequity.aspx. Accessed June 23, 2020.
      11. Levine B, Chen L, Grecu A. Achieving pay equity: how analytics has evolved to support true progress. www.imercer.com/content/common/images/knowledge-center/pdfs/achieving-pay-equity.pdf. Accessed June 23, 2020.
        12. Bloom N, Ohlmacher S, Tello-Trillo C, Wallshog M. Research: better-managed companies pay employees more equally. https://hbr.org/2019/03/research-better-managed-companies-pay-employees-more-equally. Accessed June 23, 2020.
        13. Anderson D, Bjaratóttir MV, Dezso C, Ross DG. Why companies' attempts to close the gender pay gap often fail. https://hbr.org/2019/01/why-companies-attempts-to-close-the-gender-pay-gap-often-fail. Accessed June 23, 2020.
        14. Hayes SN, Noseworthy JH, Farrugia G. A structured compensation plan results in equitable physician compensation: a single-center analysis. Mayo Clin Proc. 2020;95(1):35–43.
        Keywords:

        physician assistant; PA; compensation; wage gap; sex disparities; inequity

        Copyright © 2020 American Academy of Physician Assistants