Table 2.
Model-based Marginal Proportions for Statistically Significant Effects of Primary Predictors on Misbehavior Outcomes*
| Effects and interactions | Neglect or carelessness, % | The 10 most serious misbehaviors, % | Misconduct, %† | Careless, inappropriate peer review, % | Circumvention of human-subjects requirements, % |
|---|---|---|---|---|---|
| Main effects | |||||
| External funding of ≥20% expected by administration | |||||
| No | NA | NA | -- | -- | 5.6 |
| Yes | NA | NA | -- | -- | 2.0 |
| Federal funding held | |||||
| No | NA | NA | -- | 3.9 | -- |
| Yes | NA | NA | -- | 9.6 | -- |
| Private industry Involvement | |||||
| No | -- | 21.5 | 7.1 | -- | -- |
| Yes | -- | 28.5 | 12.2 | -- | -- |
| Interactions | |||||
| External funding not expected by administration | |||||
| Federal funding | |||||
| •No | 48.3 | 21.5 | -- | -- | -- |
| •Yes | 70.4 | 33.2 | -- | -- | -- |
| External funding expected | |||||
| Federal funding | |||||
| •No | 68.7 | 29.9 | -- | -- | -- |
| •Yes | 70.2 | 29.9 | -- | -- | -- |
*
NA, interpretation was not appropriate because of the presence of a significant interaction. All models were adjusted for sex, number of years since receipt of highest degree, tenure status, field of study, and minority status.
†
Misconduct consists of fabrication of data, falsification of data, and plagiarism.