132 Chapter 4 Table A2. Two-stage least squares (2SLS) regression estimates for different specifications on the outcome admission to supported housing Main model Different set of controls used in the 2SLS with preferred measure of leniency (residuals) Minimum number of cases by assessor † Including reassessments‡ Unadjusted leniency (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Effects of eligibility on supported housing admission (se) 0.316*** 0.328*** 0.349*** 0.333*** 0.333*** 0.305*** 0.346*** 0.343*** 0.301*** 0.319*** (0.057) (0.067) (0.560) (0.055) (0.055) (0.057) (0.049) (0.077) (0.055) (0.068) First-stage (se) Effect of leniency on eligibility 0.978*** 1.132*** 0.993*** 0.980*** 0.976*** 0.965*** 0.971*** 0.946*** 0.958*** 0.790*** (0.048) (0.123) (0.048) (0.048) (0.048) (0.049) (0.040) (0.084) (0.048) (0.054) Controlling for: Region, period and type of assessment x x x x x x x x x Patient characteristics x x x (age dummies) x x x x x Application characteristics x x x x x x Diagnosis and care requested x Minimum number of cases/ assessor 30 30 30 30 30 30 15 45 30 30 Observations 7,953 7,953 7,953 7,953 7,953 7,953 9,362 6,633 11,459 7,953 F-statistic (p-value) for weak identification (Cragg-Donald) 415.35 (0.000) 84.8 (0.000) 429.4 (0.000) 421.7 (0.000) 416.4 (0.000) 392.9 (0.000) 591.3 (0.000) 126.0 (0.000) 404.0 (0.000) 211.8 (0.000) Notes: Robust standard errors in parentheses, clustered at the assessor level; *** p<0.01, ** p<0.05, * p<0.1. † minimum number of cases by assessor, region and type of assessment ‡Applications with a valid eligibility decision to supported housing in the 365 days before the application. Results for all other outcomes are available per request for robustness checks done in columns (6) to (10): estimates are consistent with our main findings.
RkJQdWJsaXNoZXIy MjY0ODMw