Interpreting model coefficients from regression analysis

Abstract

current analysis, it is not clear that the number of operators and/or the presence of pathology predicts scan time, a subtle but essential difference. This may be the case but is not shown by the statistical analysis used. The authors would benefit by reassessing the types of variables used in the study and analysing their data using the GLM function in SPSS, which can investigate the actual effect between groups. The interpretation would be slightly different, as the output of this type of analysis provides coefficients with respect to a reference group (dummy variable). That is, the difference in scan time compared to one operator or the absence of pathology. So reporting the finding would follow something like ‘ when two operators attended the scan, on average, scan time increased by XX mins (95% CI: X (cid:1) X mins) compared to a single operator, after adjusting for the presence of pathology. ’ I hope this letter is of some assistance to the authors.