Table 3.

Matching Results on the Effect of Marriage on Recidivism, by Spousal Criminality

Model description		Model 1a	Model 2b	Model 3c	Model 4d
	Main analysis
Difference in recidivism, matched samples	Estimate	−.02***	−.02***	.03	.11***
	N (all / treated)	93,264 / 9,574e	93,264 / 9,265f	93,264 / 309	9,231g / 309
Difference in recidivism,unmatched samples	Estimate	−.08***	−.09***	.13***	.22***
	N (all / treated)	93,264 / 9,575	93,264 / 9,266	93,264 / 309	9,266 / 309
	Sensitivity analysis
Spousal criminality observed 0–2 years prior to marriage	Estimate	−.02***	−.02***	.06	.08*
	N (all / treated)	93,264 / 9,574	93,264 / 9,419	93,264 / 156	9,384 / 156
Spousal criminality observed 0–8 years prior to marriage	Estimate	−.02***	−.02***	.02	.06***
	N (all / treated)	93,264 / 9,574	93,264 / 9,073	93,264 / 502	9,073 / 502

^aIn Model 1 the treatment is marriage (to any spouse) and the control is nonmarriage.

^bIn Model 2 the treatment is marriage to a nonconvicted spouse and the control is nonmarriage.

^cIn Model 3 the treatment is marriage to a convicted spouse and the control is nonmarriage.

^dIn Model 4 the treatment is marriage to a convicted spouse and the control is marriage to a nonconvicted spouse.

^eOne observation is off common support and thus not included in the analysis.

^fOne observation is off common support and thus not included in the analysis.

^gThirty‐five observations are off common support and thus not included in the analysis.

^* p < .05.

^*** p < .001.