111 The effects of supported housing for individuals with mental disorders Table 1. Descriptive statistics: group means for total population and by supported housing eligibility (continued) Total mean (sd) Eligible mean (sd) Non-eligible mean (sd) Difference: noneligible vs eligible p-value t-test Institution for the Disabled 0.00 0.00 0.02 0.02 0.000 Negative decision 0.06 0.00 0.41 0.41 0.000 Any home care*** 0.08 0.01 0.53 0.52 0.000 Nursing 0.01 0.00 0.03 0.03 0.000 Individual guidance 0.08 0.01 0.52 0.51 0.000 Group guidance 0.02 0.00 0.10 0.09 0.000 Observations 7,953 6,818 1,135 Notes: sd – standard deviation *Prior refers to the calendar year before the year of application. #First-generation migrants are classified according to their country of birth. Second-generation migrants are classified according to their mother´s country of birth – if that is the Netherlands the father´s country of birth is considered. §Only available for those who receive specialist mental health care in the past 365 years. ¶ Including the candidate herself. ** Except for supported housing as all those eligible for supported housing in the 365 days ahead of the application were excluded. ***The rows below refer to the different types of home care granted in our population. Individuals can be eligible for more than one type of home care simultaneously. Individuals who were eligible for supported housing differ from those considered as noneligible (Table 1). Most importantly, they are older, have a lower prior personal income, are less likely to have worked, and more likely to have used specialist mental health care or home care before, or to have a diagnosis of psychotic disorder or of a disorder diagnosed in the childhood. Eligible individuals also have much higher health care expenditures in the prior year. Some of the application characteristics also differ by eligibility: applications granted eligibility were more often done by long-term care providers and were more often assessed through an extended procedure. EMPIRICAL APPROACH We estimate the effects of supported housing eligibility on a range of outcomes including long-term and health care use, mortality, employment and income, and parental outcomes. This relationship is given by (1) yi =α0 +α1Eligibilit yi +Cont rol si′ α C+v i Where yi is the outcome of interest for each application to supported housing i, Eligibilit yi corresponds to a binary indicator that takes the value 1 when the assessor decides that the applicant is eligible to receive supported housing, Controlsi refers to a set of covariates related to the applicant and to the application, and vi is an error term. is likely biased because assessors grant eligibility based on characteristics that are partly unobserved to the researchers (e.g. unobserved dimensions of health, wellbeing, behavior, social context). 4
RkJQdWJsaXNoZXIy MjY0ODMw