Table 2
Proposed rules of thumb for assessing small-study effects’ magnitudes
| Measure | Database | Outcome | Measure range and proportion of Cochrane meta-analyses contained in the range among the full/reduced database with binary/non-binary outcomes | |||
| Unimportant small-study effects | Moderate small-study effects | Substantial small-study effects | Considerable small-study effects | |||
|
|
|
|
| ||
| Full | Binary | 39.6% | 31.2% | 34.3% | 18.7% | |
| Reduced | Binary | 38.5% | 31.1% | 35.4% | 18.7% | |
| Full | Non-binary | 28.5% | 24.7% | 34.0% | 32.8% | |
| Reduced | Non-binary | 26.7% | 24.1% | 35.5% | 33.9% | |
|
|
|
|
| ||
| Full | Binary | 32.0% | 35.3% | 30.7% | 18.8% | |
| Reduced | Binary | 32.2% | 32.9% | 29.9% | 21.0% | |
| Full | Non-binary | 28.8% | 31.1% | 30.2% | 25.7% | |
| Reduced | Non-binary | 26.8% | 29.8% | 30.8% | 27.9% | |
|
|
|
|
| ||
| Full | Binary | 38.8% | 30.2% | 32.2% | 21.8% | |
| Reduced | Binary | 37.0% | 29.9% | 33.4% | 22.8% | |
| Full | Non-binary | 27.8% | 23.3% | 29.9% | 38.2% | |
| Reduced | Non-binary | 25.4% | 21.9% | 31.0% | 40.5% | |
|
|
|
|
| ||
| Full | Binary | 32.3% | 35.2% | 30.7% | 18.7% | |
| Reduced | Binary | 33.0% | 32.8% | 30.1% | 19.9% | |
| Full | Non-binary | 29.9% | 31.7% | 30.5% | 24.2% | |
| Reduced | Non-binary | 27.8% | 31.5% | 31.4% | 26.2% | |
|
|
|
|
| ||
| Full | Binary | 33.2% | 31.1% | 26.0% | 24.6% | |
| Reduced | Binary | 31.8% | 31.9% | 25.8% | 23.2% | |
| Full | Non-binary | 45.7% | 25.3% | 22.3% | 19.6% | |
| Reduced | Non-binary | 42.9% | 28.0% | 23.9% | 17.7% | |
|
|
|
|
| ||
| Full | Binary | 34.0% | 28.8% | 26.3% | 17.3% | |
| Reduced | Binary | 32.0% | 30.4% | 28.5% | 16.1% | |
| Full | Non-binary | 47.2% | 26.2% | 21.7% | 10.5% | |
| Reduced | Non-binary | 44.2% | 29.1% | 21.9% | 10.6% | |