Blog of the International Journal of Constitutional Law

Petersen on the use of social sciences in constitutional adjudication

In I·CON’s latest issue, Niels Petersen discusses the role of empirical assumptions in constitutional adjudication, and evaluates different strategies for using social science evidence. We have made this article freely available to I·CONnect readers, and we invite you to join the discussion of this important topic.

Click on the title to access the full-text paper:
Niels Petersen, Avoiding the common-wisdom fallacy: The role of social sciences in constitutional adjudication

Abstract: “More than one hundred years ago, the US Supreme Court began to refer to social science evidence in its judgments. However, this has not resonated with many constitutional courts outside the United States, in particular, in Continental Europe. This contribution has a twofold aim. First, it tries to show that legal reasoning in constitutional law is often based on empirical assumptions so that there is a strong need for the use of social sciences. However, constitutional courts often lack the necessary expertise to deal with empirical questions. Therefore, I will discuss three potential strategies to make use of social science evidence. Judges can interpret social facts on their own, they can afford a margin of appreciation to the legislature, or they can defer the question to social science experts. It will be argued that none of these strategies is wholly satisfactory, with the result that courts will have to employ a combination of different strategies. In order to illustrate the argument, I will discuss decisions of different jurisdictions, including the United States, Canada, Germany, and South Africa.”


One response to “Petersen on the use of social sciences in constitutional adjudication”

  1. Tom Ginsburg Avatar
    Tom Ginsburg

    I liked this paper a good deal, and am broadly in agreement with the sentiment that one cannot avoid confronting empirical issues in constitutional adjudication. Petersen also recognizes the problems of assuming a value-free social science.

    I was unclear, however, on one small part of Dr. Petersen’s argument. In critiquing the Federal Constitutional Court’s approach to smoking bans in Germany, he makes the point that the Court’s interpretation of the evidence of declining revenues was unsophisticated and not informed by basic statistical reasoning. So far so good. (For a study of a US jurisdiction which reported increased post-ban revenue see )

    Maybe it is a small point, but I didn’t understand the following bit of the argument (p 308): “It seems intuitively plausible that it is more likely that the numbers show at least a correlation between the smoking ban and the revenue development if the variation within the two groups of compared states is relatively small. A small variation means that the states are rather homogenous, and the other factors that might play a role have about the same effect in every state. However, a large variation would indicate that there is a great divergence in the extent to which other factors influence the development of the revenue in each particular state.” This logic may assume that smoking bans could not have a large impact, which is itself an open question.

    More importantly, heterogeneity can cancel out the effect of relevant differences. Suppose you have two states: one is warm and has a lot of smokers and the other is cold with few smokers. They are heterogenous. Ex ante, restaurant patronage is the same. Now a smoking ban is passed, say in the winter. Revenue in both states may decline the same amount but for different reasons. State one sees a deep decline in revenues because smokers wont go to restaurants. State two sees only a small decline attributable to smoking, but a larger one because of weather. the observed differences are the same, but they are for heterogenous reasons. As I said, a small point but important if one was to actually have the necessary data for a full analysis.

    Here is an interesting thought experiment: should all constitutional courts have statisticians or social scientists on staff to help them deal with the evidence?

Leave a Reply

Your email address will not be published. Required fields are marked *