Academic journal article Vanderbilt Law Review

On the Effective Communication of the Results of Empirical Studies, Part I*

Academic journal article Vanderbilt Law Review

On the Effective Communication of the Results of Empirical Studies, Part I*

Article excerpt

To claim that empirical work is a fundamental part of legal scholarship borders on the boring. It has been said (and documented) too many times for us to recount here; even the Association of American Law Schools ("AALS") acknowledged the increasing centrality of empirical work when it devoted its 2006 annual meeting to the topic.

Unfortunately, though, neither law professors nor their audiences are fully reaping the benefits of empirical work. The primary problem is that while legal academics are producing highquality research, they have been less effective at communicating the products of their labor. A strong devotion to tabular, rather than graphical, displays, and the discussion of "statistical significance" rather than substantive importance, are just two areas requiring improvement.

Adapting a burgeoning literature in the social and statistical sciences to the unique interests of legal scholars, we supply general suggestions for improving the communication of empirical studies. In the next installment (also to appear in the Vanderbilt Law Review), we outline more specific strategies aimed at effectively (and accessibly) presenting data and statistical results, and from these devise a set of protocols for implementation by legal publications.

I. INTRODUCTION

In an important and certainly timely article published in the W. Y. U. Law Review, Nancy C. Staudt demonstrates that, in taxpayer standing cases, judges are motivated by politics but can be constrained when the law is clear and oversight exists.1 As part of that demonstration, Professor Staudt offers an empirical analysis of the decision to grant standing to federal taxpayers-the results of which we reproduce in Table 1.2

What are we to make of this rather ominous-looking table? Professor Staudt suggests two key takeaways. First, the analysis, she reports, shows that doctrine helps explain standing decisions even when political factors are taken into account. Both legal variables ("Spending" and "Spending and Establishment Clause") are statistically significant, controlling for all other factors listed in the table. Second, she finds an important role for the politics of the plaintiff: Judges are more likely to grant standing to a liberal plaintiff, regardless of their own political leanings.

No doubt, the data support Professor Staudt's claim about the importance of politics. The asterisk on the Plaintiff Politics variable, for example, tells us that a "statistically significant" relationship exists between a plaintiffs political ideology and the decision to grant standing.3

Moreover, the positively signed coefficient (1.141, and not -1.141) conveys information about the direction of that relationship: Given Staudt's coding rules, and accounting for the other six variables in her analysis, liberal plaintiffs are significantly more likely than conservative plaintiffs to receive a favorable decision on their standing claim.

On the other hand, Table 1 is not just ominous-looking and offputting to most readers; it communicates information of little value either to its audience or author. For starters, while we know that liberal plaintiffs are more likely to be granted standing, we do not learn how much more likely. 0.2 times more likely? Two times? Or perhaps even four times? Certainly, this is the quantity of interest that matters most to readers.4 But it is not one that we can readily learn from a tabular display of coefficients; in fact, all we learn from the coefficient of 1.141 is that, controlling for all other factors, as we move from a conservative plaintiff to a liberal plaintiff, we move up about 1 on a logit scale-an esoteric statement at best.6

Moreover, even if we could calculate a "best guess" about the likelihood of standing for a plaintiff based on political ideology, Table 1 conveys no useful information about the error surrounding that guess. Suppose Professor Staudt reported that the probability of a judge granting standing to a conservative plaintiff is about 0. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.