Table 2.
Measures of heterogeneity in effects across comparisons
| Outcome | K | Q | p (Q) | I 2 | Τ 2 |
|---|---|---|---|---|---|
| Time 1 | |||||
| Behavior | 18 | 247.53 | <.001 | 93.13 | .096 |
| Behavioral intentions | 42 | 135.99 | <.001 | 69.85 | .037 |
| Anticipated regret | 20 | 8.67 | .979 | 0.000 | .000 |
| Negative affect about target behavior | 12 | 25.61 | .007 | 57.04 | .036 |
| Positive affect about target behavior | 8 | 75.54 | <.001 | 90.73 | .471 |
| Time 2 | |||||
| Behavior | 10 | 35.77 | <.001 | 74.84 | .064 |
| Behavioral intentions | 5 | 2.86 | .581 | 0.000 | .000 |
| Anticipated regret | 7 | 3.07 | .800 | 0.000 | .000 |
Q and p (Q) test whether the heterogeneity between studies is significant, which would suggest they do not share a common effect size (a nonsignificant p value may reflect a lack of statistical power). I2 a proportion that represents the ratio of true heterogeneity to total variance across the effect size estimates; Τ2 is a test of homogeneity between studies that removes the dependence on the number of studies and is in the same metric as the effect size (similar to a standard deviation).