The State's Attorney had previously tried to get the study released without success. This story in the Courant last week by Katie Melone says the defense claimed it was work product. For one government agency to use the work product doctrine to hide from another government agency a document of basic evidentiary facts (not legal reasoning or impressions) produced with taxpayer dollars is stretching that doctrine, to put it mildly.

Today's article, linked above, says, "Chief Public Defender Susan Storey said Tuesday that it was questions about the statistical significance of the report, and not conclusions that the race of the defendant did not play a role, that prompted her office to withhold its release."

Baloney. Statistical significance is something you deal with in the discussion of what the results mean. The fact that the results are not "statistically significant" is never a reason for suppressing a study.

Statistical significance is an aspect of the way statistical analysis in these types of studies has traditionally been done since the pre-WWII work of R. A. Fisher. If your research hypothesis is that A and B are correlated, you set up a null hypothesis that A and B are not correlated. The rule of thumb is that a correlation is "statistically significant" if you can reject the null hypothesis with 95% confidence. That is, the chance that the true correlation is zero and that the correlation you observed is just random variation is less than 5%. But that is only a rule of thumb, not magic. See generally, Jacob Cohen (1990), Things I have learned (so far), American Psychologist, 45(12) 1304-1312.

A study with results that do not meet the traditional criterion of "statistical significance" still tells us something, and what it tells us may be "significant" in the broader sense of being important. If the correlation between A and B is not statistically significant, we can calculate with some degree of confidence an upper limit of the true correlation. That is, if the correlation coefficient had been greater than X there is a 95% percent chance we would have seen it, so it very likely somewhere between 0 and X. If "less than X" means that B is at most a minor factor in determining A, that is a "significant" result in the broader sense. Statisticians, criminologists, and even lawyers can argue about the significance of facts, but that disagreement is no basis for suppressing the facts.

The methodology of the study might also be questioned. After reading it, I have some serious doubts. Again, though, criticisms of that type would be a reason to release your criticisms along with the study, not a reason to suppress the study.

The Chief Public Defender has some serious explaining to do. What we've heard so far doesn't cut it.

On the underlying question of "disparity," what we see in this study is what we've seen before. Race-of-defendant bias is either zero or so low as to not be a major factor, and this is not the first time the opponents' own study has shown that. There is a claim of so-called race-of-victim bias, meaning the death penalty is not applied often enough in black-victim cases, but that is not an argument for overturning the well-deserved sentences of those who have been sentenced already, as I have explained before. So-called "geographic disparity" is simply not a problem. As Judge David Baime said in the New Jersey disparity litigation, the various jurisdictions within a state are not supposed to "march in lock-step."

This report is not a bombshell, but its suppression reinforces the impression that the opposition regards truth as expendable in its crusade against the death penalty.