Robyn M. Dawes - US grants

Parole boards must ask themselves a crucial question: should the prisoner before them be released, or is he or she likely to commit another crime? While statistical prediction rules (SPRs) have been shown to be the best way of combining information about such prisoners to predict recidivism, parole officers tend to rely on their own intuition for combining the information to make their decisions. College admissions officers make similar sorts of errors each year, by basing admissions on criteria such as one-on-one interviews, instead of statistical models that can predict success in college based on past performance, often incremented by tests that have been specifically designed to predict performance. Nothing predicts without any inaccuracy, but the conclusion that SPRs are more accurate than intuition is well established, by over 150 studies comparing the two prediction modes. The proposed project asks a related question, one with important public policy implications: Why do people believe that they can make better decisions than can statistical models? One hypothesis to be investigated is that while SPRs automatically specify how poorly they predict when they specify how well, people overestimate their own ability to predict outcomes and hence reject the statistical models. Moreover, such overestimation may be based in part at least on the cognitive factor that our belief in our ability to predict is partially based on our ability to "explain" ("fit") situations in retrospect ("I could have told you he'd kill again, given his behavior as a child"). By understanding these biases of overestimation and retrospection, we can begin to devise effective methods for convincing people to rely more on statistical prediction rules and less on their own judgment, thereby befitting clients to whom accuracy of prediction is important, often vital.

This project will study determinants of the nation's generosity towards Hurricane Katrina victims. While media reports have been rife with allegations that the nation would have responded more strongly if these victims had not come to a large extent from minority groups and economically disadvantaged backgrounds, there is no scientific evidence to rebut or support these allegations. More generally, does the public's generosity towards Katrina victims depend on the needs, behaviors and characteristics of these victims? This study will answer these questions using a randomized experiment on a nationally representative sample in which individuals' generosity towards Katrina victims is measured by giving them the opportunity to divide a sum of money between themselves and a charitable organization helping Katrina victims. Prior to measuring their generosity to Katrina victims, the experiment will elicit beliefs about Katrina victims by asking a set of questions that have objective answers, such as the racial composition of the victims, the amount of federal aid given to each victim, or the fraction of victims who owned cars that they could have used to evacuate themselves before the storm. Next, the correct answers to a randomly selected subset of these questions will be provided, thus changing some of the subjects' beliefs. These changes in beliefs allow one to estimate the causal effect of the victims' characteristics, needs or behaviors on the generosity of support.
The scientific impact of this project is fundamental because researchers only have a limited understanding of determinants of actual redistribution in a nationally representative sample. Most of what is known comes from survey data containing self-reported preferences for redistribution, voting behaviors or charitable activities. However, self-reported data on even the simplest objective questions (e.g., how one voted in the last election) are highly prone to systematic error. To address this, a handful of studies have been conducted on the determinants of giving to real charities and real poor people in laboratory experiments. However, these studies have obvious limitations as well such as the fact that they are not conducted on nationally representative samples. There is also great methodological value to the proposed study because it will illuminate how findings from survey data and laboratory experiments replicate when real behavior is measured in a nationally representative sample. Finally, this study has a broader impact. Our nation has become increasingly concerned about how to respond to disasters as threats of terrorism have increased. There are two classes of response to such threats: prevention and preparation beforehand and responses afterward. This study will deepen our understanding of how and why Americans respond after a man-made or natural disaster has struck.