Table 3.
Predictors of Specialty Choice for Residents (Weighted Percentages)
| Distribution of Sample* | 1994, % Choosing† | 1997, % Choosing† | Fitted Odds Ratios (95% CI)‡ | |||||||
|---|---|---|---|---|---|---|---|---|---|---|
| 1994 (N = 635) | 1997 (N = 483) | Primary Care (N = 255) | IMPS or Undecided (N = 142) | Specialty (N = 238) | Primary Care (N = 198) | IMPS or Undecided (N = 82) | Specialty (N = 200) | Main Effect of Predictor§ | Interaction with Time (0 = 1994; 1 = 1997)‖ | |
| Role model | ||||||||||
| Primary care | 23.5 | 28.3 | 71.0 | 13.5 | 15.5 | 81.8 | 10.4 | 7.8 | 17.97 (11.21 to 28.80)# | 2.43 (1.17 to 5.03) |
| Specialist | 54.0 | 57.8 | 10.2 | 16.3 | 73.6 | 8.2 | 13.6 | 78.2 | P < .0001 | P < .05 |
| No role model | 22.4 | 13.8 | 21.8 | 18.9 | 59.3 | 21.1 | 9.7 | 69.2 | 9.03 (5.29 to 15.43)** | 2.69 (1.03 to 7.03) |
| P < .0001 | P < .05 | |||||||||
| Peer encouragement | ||||||||||
| Toward primary care | 23.6 | 29.1 | 58.3 | 7.0 | 34.8 | 69.1 | 13.4 | 17.5 | 5.44 (3.32 to 8.91)†† | 3.06 (1.49 to 6.27) |
| Toward specialties | 57.7 | 52.6 | 16.0 | 16.8 | 67.3 | 11.0 | 11.8 | 77.2 | P < .0001 | P < .01 |
| Toward neither | 17.9 | 17.0 | 22.3 | 27.4 | 50.3 | 28.7 | 11.3 | 60.0 | 3.01 (1.73 to 5.23)‡‡ | NS |
| P < .0001 | ||||||||||
| Faculty encouragement | ||||||||||
| Toward primary care | 41.8 | 43.6 | 35.0 | 11.9 | 53.1 | 48.2 | 10.0 | 41.9 | 1.35 (0.95 to 1.93)†† | 2.91 (1.63 to 5.18) |
| Toward specialties | 47.9 | 47.0 | 22.5 | 20.8 | 56.7 | 15.1 | 14.5 | 70.4 | P = .10 | P < .001 |
| Toward neither | 9.7 | 8.9 | 15.1 | 13.1 | 71.8 | 30.9 | 9.2 | 59.8 | 2.47 (1.28 to 4.76)‡‡ | NS |
| P < .01 | ||||||||||
| Socioemotional orientation | ||||||||||
| Socioemotional | 40.0 | 38.2 | 40.6 | 13.2 | 46.2 | 54.0 | 9.7 | 36.4 | 2.33 (1.63 to 3.32) | 2.10 (1.18 to 3.75) |
| Non-socioemotional§ | 60.0 | 61.8 | 18.1 | 18.2 | 63.7 | 16.5 | 13.7 | 69.8 | P < .0001 | P < .01 |
| Gender | ||||||||||
| Female | 35.8 | 38.9 | 35.8 | 18.6 | 45.6 | 40.7 | 9.8 | 49.5 | 1.91 (1.4 to 2.5) | NS |
| Male | 64.2 | 61.1 | 22.4 | 14.8 | 62.9 | 24.5 | 13.7 | 61.8 | P < .0001 | |
Column percentages: distribution of predictor in total population (column percents add to 100).
Row percentages (row percents add to 100).
Fitted odds ratios represent the estimated cumulative odds of choosing a more primary care–oriented career (rather than less oriented toward primary care, such as primary care rather than IMPS, undecided between primary care and IMPS, and specialties) across specified levels of predictor variables (e.g., having a primary care versus specialist role model).
In the presence of an interaction with time, a main effect indicates the effect of the predictor on career choice in 1994. Main effects for predictors in 1997 are not shown.
An interaction effect indicates that the effect of the predictor changed over time, and the estimated odds ratio represents the factor by which the odds increased or decreased.
Primary care versus specialist role model.
Primary care versus no role model.
Encouragement toward primary care rather than specialties.
Encouragement toward primary care rather than toward neither.
Includes technical-scientific (N = 610) and missing (N = 17).
IMPS, internal medicine/pediatrics subspecialty; undecided, undecided between primary care and IMPS; 95% CI, 95% confidence interval; NS, not significant.