Abstract
The kindling hypothesis for depression predicts that with more recurrences, the interval between successive recurrences decreases. Studies with unipolar and bipolar samples generally have been consistent with this premise. However, previous research is subject to a statistical artifact. Slater's fallacy maintains that these intermorbid intervals appear to decrease because highly recurrent individuals with consistently shorter intervals become a larger proportion of the remaining sample with each recurrence. Correcting for this bias, research on bipolar disorder no longer evidences such an effect. We predicted similar results for unipolar depression when correcting for this bias, and proposed an alternative model: individuals who are highly recurrent have consistently shorter intermorbid periods, even following the very first lifetime episode. As predicted, correcting for Slater's fallacy removed the appearance of decreasing intermorbid intervals. Further, highly recurrent individuals exhibited shorter intermorbid intervals in general, and for the very first interval, supporting the alternative model.