Development and evaluation of an internationally comparable scale of student socioeconomic status using survey data from TIMSS / Henry May.
xiii, 148 p. : ill. ; 29 cm.
- Local subjects:
- Penn dissertations -- Education.
Education -- Penn dissertations.
- The Third International Mathematics and Science Study (TIMSS), sponsored by the International Association for the Evaluation of Educational Achievement (IEA), was first conducted in 1995 in over 40 nations. Achievement tests were administered to over one-half million students around the world along with questionnaires for those students, their teachers, and the principals of their schools. Much of the data were made publicly available in 1998 and many researchers have performed secondary analysis of the data. To date, however, almost no work has been done to validate the measures of student socioeconomic status (SES) collected as part of TIMSS. Traditional measures of SES are typically derived from one or more indicators of educational status, occupational status, and income or wealth. TIMSS researchers collected information about both the educational attainments of each student's parents and his or her family's wealth. They did not, however, measure wealth directly. The information pertaining to family wealth consists of a collection of survey items inquiring about possessions in the home (e.g., number of books, computer, video camera, etc.). Students from higher wealth families were expected to have more of these items in their homes. In this dissertation research, a new method is presented and implemented for deriving a scale of SES from these survey items using a multilevel Bayesian Item Response Theory (IRT) model. This new model was found to be superior to previously utilized models for two reasons. First, the resultant scores were internationally comparable; that is, students with equivalent scores but different nationalities had the same SES relative to an international benchmark. Second, the model was able to reduce the influence of national differences in culture that affect survey responses, despite the irrelevance of these cultural differences to national differences in SES. The results of this research suggested that the new scaling model is far superior to a more conventional model in terms of its fit to the data and its ability to produce a valid measure of SES.
- Supervisor: Erling E. Boe.
Thesis (Ph.D. in Education) -- University of Pennsylvania, 2002.
Includes bibliographical references.
- Local notes:
- University Microfilms order no.: 3073031.
- Boe, Erling E., advisor.
University of Pennsylvania.
|Location||Notes||Your Loan Policy|
|Description||Status||Barcode||Your Loan Policy|