Behavioralists and Public Policy
June 2nd, 2009Posted by: admin
This short SEED Magazine piece describes some of the efforts to incorporate behavioral science into the design of user interfaces. The examples focus on government programs and websites. One of the recent waves of popular science books has focused on such shaping of public action, including Nudge. One of the co-authors of Nudge, Cass Sunstein, has been nominated to head the Office of Information and Regulatory Affairs in the Office of Management and Budget, so it’s possible this kind of behavioral thinking could further influence regulation.
While the science behind some of this design and thinking is new, it’s important to note that this is a more explicit treatment of the kind of behavior shaping that law and policy do all the time – limiting choice or designing systems to privilege particular outcomes or products. What makes things different now is a greater ability to design things in advance with some notion of how people will react. This makes discussion and debate over these issues more important before critical decisions were made, rather than trying to revisit them later.
June 6th, 2009 at 1:13 pm
The Office of Information and Regulatory Affairs, which Prof. Sunstein (eventually) will be confirmed to lead, was established in 1981 by the Paperwork Reduction Act. Its primary functions are to reduce the burden of government paperwork (hence the catchy name) and to rationalize federal information policy (something few people know about). That, in turn, has led to progressive demands by OIRA for improvement in the quality of information government agencies disseminate.
Information quality is a serious problem. It has been widely noted that some political appointees are inclined to interfere with science. It is less often noted, but noted in Prometheus far more often than anywhere else, that many government scientists are inclined to make policy under the cover of science. When political appointees attempt to restrain government scientists from interfering with officials’ authority to make policy, government scientists (and their NGO allies) routinely accuse them of interfering with science. Thus, it requires a very careful examination of the facts in each case to ascertain who is interfering with whom.
The remedy for this problem is a strict separation between science and policy. The NAS Red Book (1983) strongly argued against separation and in favor of a “clear conceptual distinction.” The NAS model has failed.
Strict separation is now, arguably, the law of the land. A 2000 amendment to the Paperwork Reduction Act (and thus implemented by OIRA) requires agencies to ensure that significant information they disseminate is “accurate, reliable, and unbiased,” and “presented in an accurate, clear, complete, and unbiased manner.” Agencies also must have administrative mechanisms whereby the public can “seek and obtain” the correction of information that does not meet these standards. Several hundred petitions have been submitted since 2002. In a paper presented last December, I showed that agencies have with rare exception failed to even respond to these petitions in a timely manner.
The law does not have any explicit enforcement provisions, either delegated to OIRA or in the form of judicial review. This explains why agencies have been derelict. Moreover, the courts have thus far been reticent to read any enforcement language into the law. Eventually that will change, probably when an activist court encounters a sympathetic plaintiff whose case involves politically charged facts.
OIRA also reviews “significant” draft proposed and final regulations before they are published or promulgated, respectively. Sunstein will follow a long line of Administrators of this office, each with ambitious plans to influence regulatory policy. It’s a safe bet that few readers can identify a single regulatory policy that was materially influenced by a previous OIRA Administrator from 1981-2009. Their capacity to influence regulatory policy turns out to be very limited. OIRA review occurs after all the big decisions have been made. Administrators can give all the advice they want, and agencies can pay absolutely no attention to it.
For those who appreciate the idea of accounting for behavioral science in regulatory design, the career OIRA staff have been attempting to do this for decades. They have been constrained by the limited power of their political bosses more than their bosses’ party affiliations.
June 6th, 2009 at 3:46 pm
Thanks for this detailed comment shedding additional light on the situation with OIRA. For the record, I can name an OIRA Adminstrator – John Graham, but have no idea if he materially influenced regulatory policy.
What also isn’t clear to me is exactly how the amendment to the Paperwork Act that you mention would achieve the strict separation of science and policy (do you mean politics?) you desire, even if the requirements were followed in a timely fashion. Don’t get me wrong, I have no objection to the requirements. I just don’t think they separate science from policy (or politics). I don’t think anything can, and would prefer processes that help make clear the distinctions between what science can say about the world and how that information can and cannot be used politically.
June 8th, 2009 at 6:35 am
David, first please note that I’ve fixed the profile feature that failed to give my name. Previously it was uneditable, but I followed normal IT protocol (signed out, signed in) and what previously was impossible became simple. Perhaps it is a useful metaphor for the problem of separating science and policy.
The amendment, colloquially referred to as the Information Quality Act, was a snippet of language in the FY 2001 appropriations bill for OMB and other agencies. The law contained a directive to OMB with an October 1, 2002 deadline. The law delegated to OMB/OIRA the authority to issue guidelines or regulations directly agencies to improve the “quality,” “utility,” “integrity,” and “objectivity” of the information they disseminated. Congress defined none of these terms.
OMB implemented this directive aggressively in certain substantive respects, but timidly in others (specifically its decision to implement it by guidance rather than regulation). You can read the OMB/OIRA guidelines at 67 FR 8452. I have blogged on information quality numerous times at neutralsource.org, mostly showing how once you start thinking carefully about information quality, you start seeing information quality problems everywhere.
IQA principles apply to “information” (a term that is defined in the Paperwork Reduction Act), but not to policy or to the prudential judgments of agency officials authorized to make them. That distinction alone establishes a fundamental distinction between science (i.e., “any communication or representation of knowledge such as facts or data”) and policy or judgment (“opinions, where the agency’s presentation makes it clear that what is being offered is someone’s opinion rather than fact or the agency’s views”).
Right here we can see a departure from the NAS Red Book model, which sought not separation but a “clear conceptual distinction” between science (“risk assessors”) and policy (“risk managers”). Then-EPA Administrator Ruckelshaus implemented the NAS recommendation exactly the way the NAS committee recommended against, trying to keep them separate. The NAS panel members had the best of intentions, but they did not take account of the way the EPA bureaucracy actually worked, nor did they keep in mind the fact that Ruckelshaus was brought in after a significant scandal at EPA — a scandal that could be overcome only by strictly separating the two functions. Over time, what has happened is that the “science” side at EPA has gradually but aggressively claimed jurisdiction over policymaking, and it has done so by converting as many policy decisions into nominally scientific ones. Efforts by political appointees, in whom is vested the statutory authority to make decisions, are routinely met by accusations that they are interfering with science. Political appointees back down quickly, especially Republican appointees for perhaps obvious reasons. So what they end up doing is trying to reform the Agency’s internal procedures so that the bureaucracy’s ability to arrogate decision making authority is more constrained. These efforts have not been terribly effective. Political appointees are transients; the bureaucracy lives forever.
The key IQA principle that applies here is “objectivity,” which I mentioned in my earlier reply. My heuristic rule for “objectivity” is that when government disseminates science, it must be policy-neutral: One must not be able to discern any policy preferences. This may not be able to achieve perfectly, but it is an appropriate goal if we want to simultaneously keep policy officials out of science and scientists out of trying to make back-door policy.
Objectivity (or policy neutrality) also provides a convenient null hypothesis for the evaluation of allegations that certain government-disseminated information does not adhere to the standard. OMB’s guidance for error correction requires petitioners to bear the burden of refuting this null hypothesis. Where OMB’s guidance falls short is it does not establish an evidentiary standard. It just says the case must be “persuasive,” a term that offers no help at all. The way agencies have interpreted this, a petitioner’s case that the information disseminated by the agency is not objective must be persuasive to … the agency itself. I can’t say that the guidelines were designed to fail, but it is easy to say that they were not designed to succeed.
My point was not that OMB’s existing guidelines “solve” the problem. Rather, the law authorizing them provided the tools to solve the problem. The law clearly permits OMB to establish credible enforcement tools. The Bush administration in 2001 did not do so, I think because it succumbed to a Poli Sci 101 myth that is pervasive among new administrations, and is visible today in the Obama administration. That myth is we, the new officials, are much smarter and more capable than our Neanderthal predecessors with respect to our ability to get Executive branch bureaucracies to execute the policies and decisions we make. Policy officials are more effective only when they agree with the policy views of the bureaucracies they nominally lead.
June 8th, 2009 at 6:59 am
Thanks again. I missed some nuance in your previous comments and appreciate the education.