The Texas Supreme Court just decided Merck v. Garza. The relatively short opinion rolls along (1) reaffirming Havner; (2) apparently adding the further requirement of a second well done epidemiological study "statistically significant at the 95% confidence level" that shows a doubling of risk; (3) rejecting the "totality of the evidence" ipse dixit of plaintiff’s expert; but then suddenly (4) utterly confounding us by holding "when parties attempt to prove general causation using epidemiological evidence, a threshold requirement of reliability is that the evidence demonstrate a statistically significant doubling of the risk". What?!
The whole purpose of the "doubling of the risk" requirement had been, we thought, to ensure that when a plaintiff has nothing but probabilistic evidence such evidence must actually support a "more likely than not" causal inference as to her specific illness. There are numerous agents that produce small effects (i.e. relative risks less than 2.0) and which are nevertheless unquestionably causative of human disease. Hopefully, the court meant "specific" where it wrote "general" regarding risk doubling.
Yet there’s another problem on the very next page. Apparently courts are now to "examine the design and execution of epidemiological studies using factors like the Bradford Hill criteria to reveal any biases that might have skewed the results of the study." Again: What?! We thought (and we’re pretty sure we’re right) that Hill’s list of factors were his way of assessing a given claim of general causation. And anyway, that’s not how you look for bias. This is how you look for bias: "Excess Significance Bias in the Literature on Brain Volume Abnormalities".
In sum we liked, of course, the court’s conclusion that when each piece of plaintiff’s supposedly supportive evidence is flawed "a plaintiff cannot prove causation by presenting different types of unreliable evidence." Yet, recognizing that causal inference is hard (nearly maddening sometimes) and that statistical inference is complicated and counterintuitive, we wish the court had done a better job on this one. The deviations from standard analysis will only support those who complain that the current court is "merely results oriented".