[jrubinovitz]

This is the actual paper documenting the study the article refers to: http://www.sciencedirect.com/science/article/pii/S0165032713...

"Method A total of 62 participants suffering from depression were randomly assigned to the therapist-supported internet-based intervention group (n=32) and to the face-to-face intervention (n=30)."

"Limitations Due to the small sample size, it will be important to evaluate these outcomes in adequately-powered trials."

"Conclusions This study shows that an internet-based intervention for depression is equally beneficial to regular face-to-face therapy. However, more long term efficacy, indicated by continued symptom reduction three months after treatment, could be only be found for the online group. "

They go in depth into the limitations of the study in section 4.1, but in a nutshell the sample size was too small and heterogeneous. I look forward to seeing more research in this area, but one cannot make the claim "Psychotherapy via Internet found as good as or better than face-to-face" just yet.

[duaneb]

> one cannot make the claim "Psychotherapy via Internet found as good as or better than face-to-face" just yet.

Actually, they did find it as good or better, just not conclusively.... not that one ever could.

(I agree with the analysis—a larger sample size is definitely needed—but I get annoyed by people who discard studies entirely. This is far, far better than no study.)

[ForrestN]

I would say "found as" is a little misleading when the size is so small. Maybe "initial evidence suggests" or something would be more accurate.

[bigfudge]

I think there's actually an argument not to discuss these small pilot studies outside of the immediate community of researchers. The studies aren't powered to detect a difference, and history tells us we're more likely to be mislead than enlightened by interpreting the results of small trials.

[duaneb]

> we're more likely to be mislead than enlightened by interpreting the results of small trials

Potentially yes, but when you consider anecdotal evidence to be trials with sample size 1, this is still a large improvement.