Multiple imputation for non-response in population-based HIV health care survey: Medical Monitoring Project

Cheng-Yu Lin, The University of Texas School of Public Health

Abstract

This study assessed the associations between clinical outcomes, behavioral factors and reasons for non-response among sampled subjects in the Medical Monitoring Project survey. The main study focused on Houston adults with HIV ages>18 who were selected for the Medical Monitoring Project sampled population from 2009 to 2013. In this study, almost half of sampled persons were classified as non-respondents during this period. Large non-response rates not only result in less efficient estimates but also preclude the use of traditional complete-data methods to analyze the data. A multiple imputation technique conducted in this study to perform the analyses to overcome the large portion of missing observations. The study found there were significant differences between the respondents and non-respondents regarding variables of age at interview, self-identified sexual orientation and level of education attained. Furthermore, we found only age at interview, self-identified sexual orientation and level of education attained were statistically significant in regard to disposition status. ^

Subject Area

Biostatistics

Recommended Citation

Lin, Cheng-Yu, "Multiple imputation for non-response in population-based HIV health care survey: Medical Monitoring Project" (2016). Texas Medical Center Dissertations (via ProQuest). AAI10126218.
http://digitalcommons.library.tmc.edu/dissertations/AAI10126218

Share

COinS