Skip to main content
. 2018 May 11;33(9):1574–1581. doi: 10.1007/s11606-018-4462-2

Table 2.

Summary of Multivariable Analyses of Characteristics Associated with Physician Income

Model Male/female difference in income, coefficient (95% CI), $
Model 1: no adjustment 90,890 (27,769 to 154,011)
Model 2: state random effects + practice random effects + practice characteristics random effects 51,308 (9636 to 92,981)
Model 3: hours worked per year + state random effects + practice random effects + practice characteristics random effects 39,521 (6646 to 72,396)
Model 4: hours worked per year + specialty group + state random effects + practice random effects + practice characteristics random effects 26,535 (526 to 52,543)
Model 5: hours worked per year + percent of hours worked per category + specialty group + state random effects + practice random effects + practice characteristics random effects 24,726 (−1174 to 50,625)
Model 6: hours worked per year + percent of hours worked per category + specialty group + percent of patient care time spent performing procedures under and not under anesthesia + compensation type + state random effects + practice random effects + practice characteristics random effects 26,679 (−1048 to 54,405)
Model 7: hours worked per year + percent of hours worked per category + specialty group + percent of patient care time spent performing procedures under and not under anesthesia + compensation type + age + years in practice + race + ethnicity + state random effects + practice random effects + practice characteristics random effects 27,404 (3120 to 51,688)

All models include Winsorized income at the 95th percentile and are restricted to observations with less than the 95th percentile of annual hours worked (< 3550 h per year) to avoid the influence of outliers. They also include non-response weights and imputed independent variables to avoid the influence of non-response and missing data. Models 2 through 7 additionally include standard errors clustered by state and practice