Identification Problems in the Social Sciences

点击翻译按钮获取中文摘要

Econometricians have long found it useful to separate inferential problems into statistical and identification components. Studies of identification seek to characterize the conclusions that could be drawn if the researcher had available a sample of unlimited size. Studies of statistical inference seek to characterize the generally weaker conclusions that can be drawn given a sample of positive but finite size. Statistical and identification problems limit in distinct ways the conclusions that may be drawn in empirical research. Statistical problems are most severe when the available sample is small. Identification problems are most severe when the researcher knows little about the population under study and the sampling process yields only weak data on the population. This research continues the investigator's program of work on identification problems in the social sciences. Four projects combining methodological research and empirical analysis are planned: Nonparametric Regression When Outcomes and/or Regressors are Censored Whereas there is a voluminous literature on regression analysis when outcomes are censored, surprisingly little attention has been given to the inferential problems that arise when regressors are censored. This project examines regressor censoring from the same nonparametric perspective used earlier by the investigator to analyze outcome censoring and other non identifying sampling processes. The findings are used to characterize the possible implications for survey research of item and interview nonresponse. Monotone Treatment Response The standard formalization of the analysis of treatment effects assumes that each member of a population of interest receives one of a set of mutually exclusive and exhaustive treatments, and that the outcome under the realized treatment is observable. Outcomes under the non realized treatments are necessarily unobservable; hence these outcomes are censored. In earlier work, the investigator has studied the nonparametric identification of treatment effects in the classical case where there are two treatments and the objective is to compare distributions of outcomes under these treatments. This project examines the identifiability of treatment effects when treatments are ordered and response functions are known to be monotone or concave monotone. The findings have application to the econometric analysis of market demand and of production. Empirical Analysis of Sentencing and Recidivism The question of how judges should sentence convicted juvenile offenders has long concerned policy makers and criminologists. The investigator is using nonparametric methods to compare the effects on recidivism of the two main sentencing options available to judges: confinement in residential treatment facilities and diversion to nonresidential treatment. This project seeks to illuminate the basic ideas of nonparametric analysis of treatment effects through a serious empirical case study, to add new methodological findings, and to contribute substantively to the field of criminology. Analyses of the Survey of Economic Expectations The nature of subjective expectations is a shared concern of researchers who study human behavior. The investigator has recently designed and implemented the Survey of Economic Expectations (SEE). SEE is an ongoing national survey initiated in an effort to learn how Americans perceive their near term economic prospects. Probabilistic questions are used to elicit expectations about income, employment, health insurance coverage, and crime victimization one year into the future. Recently completed analyses of SEE data examine respondents' income expectations and their perceptions of economic insecurity. This project would continue collection of SEE data at the present rate of about 1500 interviews per year and would perform a set of special studies that should enhance the value of the survey.