Skip to main content
. 2016 Aug 13;73(7):1175–1184. doi: 10.1093/geronb/gbw096

Table 2.

Standardized Probit Regression Coefficients (β) for Mediation Model in Full-Sample and Multigroup Analyses

Full-sample analyses Selected multigroup analyses
Model 1 Model 2 Model 5 Model 7 Model 9
CS. Long. Gender Race/Ethnicity Age
Effect N = 5,250 N = 2,013 Male Female p Valuea White Minority p Valuea Young Old p Valuea
n = 2,243 n = 3,007 n = 4,446 n = 804 n = 4,318 n = 932
Frail2←Frail1 .31**
Frail2←C.Stress .05
Frail2←Ctrl. −.15*
Frail2←SES −.10
Frail2←Age .20**
Frail2←Male .01
Frail2←White −.04
Frail1←C.Stress .02 .05 .03 .00 NS .01 .17 NS −.01 −.01 NS
Frail1←Ctrl. −.16** −.02 −.10 −.20** NS −.17** .06 NS −.16* −.31** NS
Frail1←SES −.03 −.12 −.14 .05 .03 −.03 −.09 NS −.06 −.02 NS
Frail1←Age .27** .18** .27** .28** NS .28** .27* NS
Frail1←Male .03 .02 .04 .01 NS .01 .12 NS
Frail1←White −.10** −.09* −.07 −.14** NS −.11** −.07 NS
Ctrl.←C.Stress −.40** −.33** −.44** −.36** NS −.40** −.39** NS −.41** −.25** NS
Ctrl.←SES .33** .24** .33** .27** NS .32** .37 NS .34** .28 NS
Ctrl.←Age −.29** −.20** −.21** −.36** NS −.30** −.16 NS
Ctrl.←Male −.01 .03 −.00 −.15 NS −.01 .14 .05
Ctrl.←White −.03 .05 −.01 −.04 NS −.03 −.10 NS
C.Stress←SES −.10* −.07 −.19** −.01 .02 −.07 −.37* .03 −.13* .09 .02
C.Stress←Age −.15** −.10* −.14** −.14** NS −.15** −.22** NS
C.Stress←Male −.05 −.04 −.07* 11 NS −.04 −.12 NS
C.Stress←White −.07* −.01 −.10** −.03 NS −.05 −.16** .03

Notes: Results of Models 3 and 4 (cross-sectional and longitudinal moderation models) are not presented. Models 6 and 10 (longitudinal subsamples) are not presented, as results were highly similar to Models 5 and 9.

CS = cross-sectional; C.Stress = chronic stress; Ctrl = perceived control; Frail1 = frailty at baseline; Frail2 = frailty at time 2; Long = longitudinal; NS = not significant; SES = socioeconomic status.

aProbability value for critical ratio (z score) of difference between parameters across groups.

p < .05. *p < .01. **p < .001.