Table 7.
Results of generalized linear regression modelsa
| Response (dependent) variable | Number of resources provided | Number of changes caused |
|---|---|---|
| Intercept for count model | 2.82** (1.31, 5.94) | 0.14*** (0.06,0.32) |
| Gender (male vs. female) | 0.72 (0.42, 1.19) | 1.85* (1.16, 3.04) |
| Academic rankb (full vs. assistant) | 1.70* (1.08, 2.62) | 1.76* (1.03, 3.06) |
| Academic rank (associate vs. assistant) | 2.64*** (1.57, 4.98) | 1.12 (0.68, 1.88) |
| Position type (DBER vs. non-DBER) | 2.09* (1.06, 4.10) | 2.90*** (1.62, 5.24) |
| Teaching PD (participant vs. nonparticipant) | 0.98 (0.5, 1.74) | 2.03* (1.14, 3.60) |
| Survey participant (Y vs. N) | 1.90** (1.23, 3.08) | 1.43 (0.98, 2.12) |
| Total resources provided | NA | 1.15*** (1.13, 1.19) |
| Department Yellow vs. Blue | 1.17 (0.72, 2.12) | 0.58* (0.36, 0.92) |
| Department Red vs. Blue | 1.01 (0.45, 2.40) | 0.72 (0.39, 1.30) |
| Department Green vs. Blue | 1.09 (0.68, 1.89) | 0.37** (0.20, 0.68) |
| Intercept for binary model | 2.48*** (0.81, 3.68) | NA |
aEstimates and 95% confidence intervals of the association between explanatory variables and reports of teaching resources provided and change caused as a result of interactions about undergraduate teaching with departmental colleagues.
bIn these models, assistant professor and the Blue department are set as the reference levels of the academic rank variable and department variable, respectively. One can fit another model with different reference levels to determine the relationship between these levels and the response variables. We examined all of these. The only significant results were associate vs. full professors for resources provided: 0.65*(0.35, 1.11) and Red vs. Green for change caused: 1.95*(1.10, 3.49). Setting a different reference level does not change other results, except the estimate of the intercept.
*p < 0.05; ** p < 0.01; *** p < 0.001 (two-tailed tests).