Letter to the Editor: Sources of bias and need for caution in interpreting the results of Spoth et al.'s (2017) PROSPER study

[...]claims of intervention effectiveness are problematic for at least three reasons: the potential for selection bias in the sample used in the study; the low rate of participation in the SFP 10-14; and the possibility that the positive results reported were the result of flexible data analysis and selective reporting. Since 21% of subjects did not complete the 9th grade follow-up assessment and 28% did not complete the 10th grade asesssment (Spoth et al. 2011, Figure 1), it is unlikely there were no refusals when a random sample was drawn from these same subjects at the 7.5-year follow-up. There are five other variables common to each analysis, and none of the differences in scores on these between the PROSPER and control groups were statistically significant using two-tailed tests. [...]the 'pattern' of statistically significant effects Spoth et al. Each index is comprised of five substances, but only two substances are common to both indexes (methamphetamines and ecstasy). [...]what appears at first sight to be a consistent effect across the two follow-ups, is not, and no reason is given for the change in the composition of the index.