 |
 |
Well, the original message just states "But some how I got different results". Different smoothness estimates would affect related p values, leading to "different" results (at least possibly / to some extent) in the table. If one of the first 64 images were an extreme outlier, included in model 1 but not in model 2 when calculating smoothness, then the impact might be larger. Of course, it could be due to accidentally loading incorrect images when setting up the model, contrasting incorrectly and/or a naming/order issue. Best Helmut