18) [McGruff the crime dog, noted prevention scientist] I heard you left the "CPPRG". What did you do, act up one too many times?"

No, I mostly left the Conduct Problems Prevention Research Group because I was interested in working in other areas. I did have some disagreement with the group about how science best proceeds. My model of science is chaotic. For example, the debate in demography about the consequences of teen childbearing is a good example. That debate involved data that were publicly available with a frank and open discussion of what various researchers were finding. Such a debate is open but not tidy and is a lot more informative. What emerged over time was a balanced perspective on the consequences of teen childbearing and a good sense of the robustness of the findings.
What I've observed among psychologists is a desire for a very tidy story, too tidy for my taste. Maybe it grows out of the laboratory tradition, which economists haven't had. But I don't see any thing wrong with putting out a preliminary paper on some outcome that presents a standard set of analyses (such as those outlined in the grant application originally). Then, if subsequent, more exploratory analyses show different effects (larger or smaller), I don't see anything wrong with a second paper refining (or even contradicting) the first paper. I do understand that this complicates (or even hinders) the ability of researchers to have a policy impact. However, I think bad science is too high a price to pay for such an impact.
(I also think the distate for complexity among psychologists reflects their limited methodological training (meaning the applied psychologists, not the methodologists). They just don't have much interest in sorting through various parameterizations of models and so on.)
I also believe large studies have an obligation to generate such preliminary studies in a timely manner. And to make the data publicly available. Psychologists have this sense that "we worked hard to collect these data and should benefit from them". However, in most cases, those same psychologists were paid to collect those data with public funds. The data should be made available almost immediately.

I also think we have some real disagreements about how the group has proceeded over time, but hindsight is 20-20. For example, I think it's clear that FT is four studies not one, but maybe this was not apparent initially. As was pointed out to me, the grant was funded multiple times, and reviewers didn't raise this issue. That's fair enough. I do think there is a big difference between pooling samples and combining point estimates. These are all things that could be explored in that study.

I also think a systematic effort to understand why effects weren't bigger in the earlier years was an opportunity missed. For example, we now know the sample was padded with less severe kids to boost sample sizes. (I don't mean "padded" in a sinister way. I just mean--like all studies--individuals were recruited until sample size goals were met.) Who were those kids who were severe and wouldn't participate? Why couldn't those kids and their families be engaged?

I also didn't think the group was sufficiently productive, especially in the core areas of intervention impact, but I'll leave that to others to decide. I am very conscious of having five R01's tied up in one project when money is so hard to get, especially for junior people.

I have no investment in whether Fast Track works or not. I'm not like my buddy Len Bickman. I don't believe "nothing works". I really just want to know. I do know that I have benefited enormously from my involvement and am very grateful to the group.

And most of my complaints could be levelled at any study. More generally, I just think developmentalists spend way too much time doing the same things over and over and over instead of really digging deeper into new questions with new methodology.