It seems a like a different possible explanation is a variation in publication bias (and perhaps researcher bias) over time in response to the changing status o

It seems a like a different possible explanation is a variation in publication bias (and perhaps researcher bias) over time in response to the changing status of CBT. Studies saying "this new thing works" have an obvious appeal, particularly in a field searching for something beyond classic psychotherapy. So when CBT first appeared, studies with strong, positive results may have been favoured. Studies saying "the mainstream thing doesn't work", or "this mainstream thing doesn't work as well as we thought" also have obvious appeal. Now that CBT is mainstream, research showing that it is less effective that thought, or not effective in particular types of cases, may be more likely to be published.