Table 2 Mean Cronbach’s alpha calculated from 50 random samples of size N from each of the BWS and Likert conditions. Higher values of alpha indicate increased reliability for the face-level scores in that condition. BWS best-worst scaling
From: Best-worst scaling improves measurement of first impressions
N | BWS | Likert |
|---|---|---|
8 | 0.703 | 0.655 |
12 | 0.814 | 0.745 |
20 | 0.870 | 0.839 |
30 | 0.911 | 0.883 |
40 | 0.931 | 0.914 |
50 | 0.947 | 0.929 |
166 | 0.984 | 0.978 |