The project aims to solve a dilemma that researchers interested in psychological and social measurements often encounter. It can be described as a situation in which a researcher has to choose between a more complex measurement model that is less interpretable and a simpler model that, without adaptation, may suffer from misspecification. To provide solutions to this problem, the project will study the approach of using simple and interpretable models that are sufficiently good approximations of reality, and then perform statistical adjustments for the potential deviations of the simple models from empirical data. Specifically, the research focuses on a branch of contemporary measurement models (i.e., item response models) to illustrate such an approach. The research will study two methods that implement this approach. The first method treats the deviation of the model from data as a nuisance factor. This method will be applied to solve a dilemma in computerized adaptive testing for measuring self-perceived physical functioning in elderly populations. The second method, which provides substantive models for the deviation of simpler models from the data, will be applied to two other social and behavioral measurement situations. The project will deliver flexible and meaningful analytic tools for social scientists facing the interpretation-versus-misspecification problem.

The project also has potentially significant broader impacts through increasing the flexibility of researchers in measuring psychological constructs in important domains such as educational achievement, psychological well being, quality of life, and social beliefs. The results of the project will be applicable to a wide range of psychological and social measurement situations in which the misspecification-versus-interpretation dilemma is prevalent. Our society is increasingly emphasizing the need to understand and measure individual needs, whether they are social, economical, or health-related, and to provide customized solutions. Providing solutions for this dilemma will help researchers to establish interpretable and accurate metrics in various social and psychological domains. Finally, the project also will improve research infrastructure for the social and behavioral sciences by offering publicly available software tools for performing measurements.

Agency
National Science Foundation (NSF)
Institute
Division of Social and Economic Sciences (SES)
Application #
0719354
Program Officer
Cheryl L. Eavey
Project Start
Project End
Budget Start
2007-09-15
Budget End
2011-08-31
Support Year
Fiscal Year
2007
Total Cost
$281,061
Indirect Cost
Name
Wake Forest University School of Medicine
Department
Type
DUNS #
City
Winston-Salem
State
NC
Country
United States
Zip Code
27157