0
Linear fit?

It seems that the significant findings are really an artifact of the decision to perform a median split. Looking at the scatterplot, a linear relationship of SPANE-B on APS does not seem to fit well!

waiting for moderation
1 Answer
0
Accepted answer

The study was not a correlation study. The study compared two groups, and the absence of a clear correlation (or the failure to catch one with regression techniques) does not however influence the employed tests or the obtained results.

Similar studies, where groups are compared, usually report boxplots only. We included the scatterplot in Figure 4 for the sake of completeness. Indeed, one reviewer and the editor requested us to try regression analyses after they saw the picture (a quadratic regression appears to fit here more than a linear regression). However, no meaningful results could be obtained. This is reported briefly in the paper and perhaps could have been stated more clearly. The peer review history is publicly available where earlier versions and more detailed comments around this topic can be found.

waiting for moderation