Nice test coming up

A recent retraction in Society & Natural Resources should test the researchers’ objectivity. Story here.

The original paper claimed that the U.S. Environmental Protection Agency (EPA) was racist, in that it was less likely to fine companies who polluted in areas where poor or ethnic minorities lived. It was retracted on 19 Dec 2012, because a coding error – transposed zeroes and ones – had reversed some, possibly all, of the results. These things happen. Full marks to the authors for retracting, but a revised paper should appear, once the error has been corrected.  It would be unscientific to only publish if A increased B but not if A decreased B.

But this is sociology. The lead author is co-director of the Center for the Study of Social Justice at the university of Tennessee.  Here‘s the abstract of the original “retracted” paper. It’s difficult to read over the “retracted” watermark, so here is a clean version.

Despite its responsibility to handle compliance and enforcement concerns, the U.S. Environmental Protection Agency (EPA) has become more concerned with protecting the interests of those it is supposed to be regulating than with the communities affected by environmental hazards. This research adds to the debate over environmental justice and injustice by examining U.S. EPA enforcement activity in Tennessee and the relationship between a Census place’s race and income structure and the monetary fine (if any) assessed to a violating facility in that census place. Data were taken from the U.S. EPA Enforcement and Compliance History Online Database and the U.S. Census, and both logistic and truncated regression techniques were utilized to model outcomes. The results provide evidence for environmental racism with regard to whether or not a fine was assessed; however, the amount of fine assessed appears unrelated to the racial and economic structure of the places where violations occurred.”

The opening sentence suggests that the authors knew what they were going to find before they looked at the data.  But maybe I’m overly sceptical. The full text of the original (retracted) paper would provide a great test of the researchers’ objectivity. It would serve the same function as an analysis plan for a randomised trial, published before the treatment codes were broken. All the endpoints, covariates and statistical tests would be pre-specified.

If these authors really are disinterested seekers after truth we might expect the new abstract to end something like this.

“Fines were imposed in a racially discriminatory way, but in the direction of positive discrimination. Polluters in poor and minority regions were more likely to suffer fines from violations. However, the size of fine was unrelated to the racial and economic structure of the place where violations occurred. These results refute the assumption of environmental racism by the EPA against poor and minority ethnic groups.”

We’ll see.

Jim Thornton