I’ve been reading Chris Mooney’s commentary (here, here
and here) about the Office of Management and
Budget’s (OMB) guidelines for implementing the Data Quality Act. He points out that it has
been turned into an impediment to environmental regulation by industry. I’ve
looked at the example of the U.S. EPA, and as you’ll see below, there are
examples that make his point.
At the same time, I’ve been a firm adherent of EPA’s Data Quality Objectives (DQO) process for developing study designs for investigating hazardous waste sites. Investigations that are planned without DQOs inevitably resemble very expensive, failed treasure hunts. Does my insistence that waste site investigations be structured around specific questions and defined decision points put me on the side of the OMB? Not sure – but what I do know is that DQO guidance is part EPA’s long-standing Quality System, which was incorporated into its data quality guidelines developed in 2002 in response to the Data Quality Act. So, I am a user of the Data Quality Act and will say that it can have a role in producing useable data for environmental decision-making.
However, I agree that the system is open to abuse to bogging down regulatory decision making, with the provision for challenges to the data developed by EPA. For example, the Perchlorate Study Group challenge of EPA’s risk assessment for perchlorate included requests for high-resolution images of slides of brain tissue from rats dosed with perchlorate, information on conditions under which the slides were stored and prepared, down to who sliced the tissue sections and what kind of a microtome did they use. To the naïve eye, this could look like the height of obstructionism. But these requests may make some sense. According to the PSG,
“As EPA knows, it is widely believed that the neurodevelopmental effects observed by EPA were artifacts of laboratory errors. This information is essential because there is a serious danger that differences in tissue compression during the histology could have created an apparent perchlorate effect by artifact alone.”
(Hah – so there is some recognition that there could be a neurodevelopmental effect from perchlorate exposure!)
Since the animal testing data is important in developing the Reference Dose for perchlorate (the basis for the 1 ppb action level in drinking water proposed in 2002), perhaps chasing down artifacts in pathology is important. However, while there is no discussion of how important is the data from that particular animal study (I won’t debate the issue right this minute), there is the point that noone, noone collects perfect data – not the government, industry or academia. Some researchers are better than others, but noone is perfect. Without some role for weight of evidence from multiple studies and some tolerance for making decisions and drawing conclusions under uncertainty, all of science would grind to a halt.
What’s unstated in this debate over data quality in regulatory decision making is the failure to come up with a reasonable mechanism for making decisions under uncertainty. I’ve struggled with how to articulate this view, but help has recently arrived in the form of a review published in Environmental Science and Policy. I’ve only reviewed the abstract of this article (the article is available for a fee), but it sounds very promising:
“. . . mischaracterization concerning the use of science in the U.S. policy process has lead to unreasonable expectations about the role that scientific information can play in the development of environmental and public health policies. This in turn has lead to implementation of misguided and self-defeating policy initiatives designed to ensure the objectivity or "soundness" of scientific inputs to the policy process.”
The authors argue that scientific findings cannot be stripped of their social contexts, such that scientific assessment conducted in support of policy making rarely, if ever, lends itself to descriptions of “objective” or “non-objective”. Therefore, policy initiatives such as the Data Quality Act, which are intended to assure objectivity, reliability, transparency, etc. etc., are in the end, exercises in futility. Their perspective on solving the problem is a bit vague, “scientific findings draw much of their validity through the context of their application in policy narratives”, but I think the diagnosis is sound – we’re expecting too much from science in solving environmental problems.
Thanks to Prometheus for the link to the review.