Publication Type: Conference Paper/Unpublished Manuscript
Review Method: Peer Reviewed
Abstract: What are the properties of hypothesis tests in selection models? This question is typically unaddressed in the application of selection models for discrete choice data, but it has serious implications for inference in the analysis of censored data. In this paper we focus on a discussion of the widely used bivariate probit selection model. The properties of testing for selection effects are analyzed using a Monte Carlo experiment. The results indicate that the tests are highly sensitive to model specification assumptions about identification and exogeneity of the regressors and that tests for selection have very low power in
specific cases. The tests generally have incorrect size and very low power, even in what would be considered large samples. Two examples, using a new data set on EU entry decisions (Schneider 2003), and a replication of Reed (2000) are used to demonstrate the results. Finally we offer suggestions on robust methods for testing selection effects in discrete choice models, such as Bayesian methods and bootstraps.

Publication Type: Proceeding
Abstract: Studies of mixed-member systems have addressed whether the use of two sets of electoral rules creates a divergent patterns of legislative behavior. Few of these studies focus on committees, which is surprising given the fact that the debate in the U.S. Congress literature is framed around distributional theories. Distributional theories posit that due to the need to win election in local districts, members will self-select on to particular committees and will be preference outliers. If such pressures exist in mixed systems, they should only exist for those legislators elected in local districts, and not those elected in proportional representation elections. Through a statistical analysis of the budget committee of the Russian Duma, this article shows that evidence of a mandate divide was rather limited. The variation in evidence of the mandate divide appears to be caused by two factors: the percentage of single-member district legislators on the committee and the level of party cohesion.

Publication Type: Conference Paper/Unpublished Manuscript
Review Method: Peer Reviewed
Abstract: Using data from a national survey of public attitudes toward homeless people, we evaluate the applicability of the contact hypothesis to ingroup-outgroup relations that fail to meet the optimal conditions specified in the contact literature. Our research extends past efforts by (1) moving beyond face-to-face encounters to consider multiple types of ingroup exposure to a highly stigmatized outgroup, (2) examining a variety of attitudinal outcomes, and (3) incorporating community context as a possible antecedent of such outcomes. Even after taking selection and social desirability processes into account, we find that all types of exposure affect public attitudes in the predicted (favorable) direction. Moreover, the size of the local homeless population--our primary measure of context--shapes opportunities for most forms of exposure and thus influences attitudes indirectly. These findings suggest that the scope of the contact hypothesis needs to be widened rather than narrowed. They also raise a series of issues for further investigation.

Publication Type: Conference Paper/Unpublished Manuscript
Review Method: Peer Reviewed
Abstract: The 1980s were the high water mark of concern and public programming directed at (curbing) teen pregnancy rates, which have declined in the years since (Child Trends, 2002). This might lead one to conclude that participation in sex education programs was an effective deterrent to teenage premarital fertility. This paper simultaneously tests this idea with a competing hypothesis, that social capital has a greater effect on adolescent fertility than sex ed programs. The rationale is based on Coleman’s (1988) argument that social capital facilitates both the internalization and enforcement of societal norms for behavior, in this case adolescent childbearing. An important feature of social capital is what Coleman termed “closure” – where the people one knows also know each other. I find that certain forms of social capital – that between an adolescent and his/her parents and his/her school officials – have significant effects in lessening the odds of a teenage birth by the tenth grade, while neither parental relations with the school and its officials, nor sex ed participation, has any effect. While this runs counter to the closure hypothesis, these results strengthen social capital as an important concept for the process of teenage premarital fertility.

Publication Type: Conference Paper/Unpublished Manuscript
Review Method: Peer Reviewed
Abstract: Conventional wisdom holds that people with higher educational level are more likely to engage in reading newspapers, thus gaining more knowledge than those with lower educational level. However, a specific question about the difference between higher SES and lower SES people in terms of their exposure to newspaper versus exposure to television has not been directly answered through empirical test. The present study tested a basic model of the relationship between socioeconomic status (SES) and media use. Using logistic and multiple regression analyses, the findings of the present study suggest that higher SES people have a significantly higher probability of reading newspaper than lower SES people, and higher SES people spend significantly less time watching television than lower SES people.