| Mean Ratings, N = 82.
(Largest standard error is 0.16) |
|||
| good-natured | bold | humorless | |
| level-headed | 8.45 | 6.27 | 3.63 |
| unsophisticated | 5.68 | 3.54 | 1.86 |
| ungrateful | 4.35 | 2.58 | 1.47 |
| Best Fitting Additive Model Scale Values and Predictions | |||
| good-natured 4.06 |
bold 2.03 |
humorless 0.22 |
|
|
level-headed 4.02 |
8.08 | 6.05 | 4.24 |
|
unsophisticated 1.59 |
5.65 | 3.62 | 1.81 |
|
ungrateful 0.70 |
4.76 | 2.73 | 0.92 |
The figure below plots the mean ratings in the upper panel and plots the predictions of the additive model in the lower panel.
Does the model fit the data? This question is really simply asking whether there is a significant non-additive interaction of the two factors. A within-subjects (repeated measures) ANOVA yields a highly significant interaction, F(4,324) = 35.30, p<.001. Therefore the simple additive model does not fit these data.
Copyright © 1999 John K. Kruschke