Methods: This is a secondary analysis of data from a Homeless Management Information System (HMIS) intake questionnaire of 11,425 males receiving services from homeless agencies in a Southeastern state. The data are from a standardized HMIS Common Intake Form used by all agencies involved in a continuum of care. The dependent variable is reported domestic violence (dichotomous variable). Independent variables are age, race, marital status, number of reported disabilities, number of times reported homeless, level of education, and veteran status. Given large percentages of missing data on some variables, multiple imputation, maximum likelihood, and listwise deletion were used to manage missing data, and results of analyses using the different methods compared to assess possible biasing effects of missing data. Binary logistic regression is used to analyze the data.
Results: Results based on multiple imputation of missing data show age, B = -.03, Wald=1.4, p<.005; and number of times homeless, B = .11, Wald=9.5, p<.005, related with the likelihood of having experienced an episode of domestic violence. Results based on maximum likelihood imputation show age, B = -.03, Wald=32.3, p < .005; number of times homeless, B = .18, Wald(1)=26.1, p<.005; and number of disabilities, B = .06, Wald=30.5, p<.005, related with domestic violence. Results based on list-wise deletion show age, B = -.03, Wald=9.1, p<.005; and number of times homeless, B = .25, Wald=6.2, p<.02, related with domestic violence. These findings converge to suggest as age increases, the likelihood of reporting domestic violence decreases; and as the number of times a male is homeless increases, the likelihood of reporting an episode of domestic violence increases. Results also suggest the possibility that as the number of disabilities increases, the likelihood of domestic violence increases. This finding, however, may be an artifact of how missing data is handled.
Conclusions and Implications: Results converge across analyses to suggest that age and number of times homeless are associated with domestic violence against men seeking assistance in homeless agencies. The results hint that number of disabilities is also predictive of domestic violence against men. The large percentages of missing data for some variables suggest caution in interpreting these findings. These findings need to be replicated in future studies.