Transitional House Occupant Length of Stay Modeled using Generalized Estimating Equations

The homeless population in the United States has seen a dramatic change in the last three decades with an increase in women, children, and families all living without nightly shelter (National Coalition for the Homeless, 2009). The changing dynamic of the homeless population has increased pressure on the public health of communities and the social service networks that aim to provide adequate services to vulnerable populations with an increased risk of mental, physical, emotional, and health related problems (Aday, 2001). The Greeley Transitional House works to provide emergency and extended housing to homeless families of at least one adult caring for at least on dependent child, helping minimize the public health risks to the community. An important consideration in managing the Greeley Transitional House is the typical occupant length of stay. In addressing this concern, this paper presents a number of Poisson log-linear models (Agresti, 1996) using the number of days in an individual's length of stay as the response. These models are used to assess the impact on length of stay of such individual characteristics as gender, age, size of family, education, and reason for entering the transitional house. The models are also used to assess the impact of such individual treatments as case management hours and job training. To account for some individuals who participate in the Greeley Transitional House multiple times, these models are extended using Generalized Estimating Equations (Zeger and Liang, 1986) to account for the autocorrelation inherent in repeated observation of individuals.