Figure 3. Estimated Treatment Effects of Race and Gender of Simulated Physicians on Composite Index and Primary Outcome Measures.

Covariate-adjusted treatment effects from ordinary least squares regression with control group (simulated white male physician). Estimates are pooled across 2 independent patient analog experiments (N = 3215) and standardized using the Glass delta, which scales outcomes by the SD in the control group. Composite index (range, 0-100) was created by extracting the first principal component from a principal component analysis on all primary outcome measures (patient confidence, patient satisfaction, likelihood to recommend, believes symptom checker, and requests more tests) (eMethods 4 and eTable 5 in Supplement 2). Dots indicate means; cyan lines, 90% CIs; gray lines, 95% CIs; and cyan dotted vertical lines, margin of equivalence within 0.20 standard units.