. Author manuscript; available in PMC: 2011 Dec 1.

Published in final edited form as: Psychol Health Med. 2010 Dec;15(6):660–671. doi: 10.1080/13548506.2010.507772

Table 3.

Regression Models of Mediators between Beliefs about Trust and Condom Use and UAI

	X→ M	M →Y	X, M →Y	R²	Sobel Test

					z	p
Beliefs about trust and condom use (X) / UAI (Y) (X →Y: β = .163^*)
Knowledge (Mk)	−.146^*	−.149^*	.143^*	.057	1.627	.104
Attitudes (Ma)	−.199^**	−.251^***	.115	.090^**	2.506	.012
Social Norms (Msn)	−.166^**	−.155^*	.144^*	.058	1.764	.078
Self-efficacy (Mse)	−.232^***	−.112	.145^*	.045	1.509	.131
Intentions (Mi)	−.209^***	−.404^***	.091	.188^***	2.962	.003
IMB Constructs (Mi, Ma)			.082	.193^***
IMB Constructs (Mi, Ma, Msn, Mse, Mk)			.089	.200^***

Note. 1. X: Predictor; M: mediator; Y: Outcome. 2. Regression models were controlled for the variables of ‘age,’ ‘race × township,’ and ‘self-reported HIV status’ which were found to be significantly associated with predictor and mediator variables. 3. Standard regression coefficient X →Y was .163 (p = .012). 4. Sobel test statistics are calculated using the following formula: z-value = a*b/SQRT(b2*sa2 + a2*sb2), where a = unstandardized regression coefficient (B) for the association between X and M; sa = standard error (SE) of a; b = B between M and Y; sb = SE of b.

p < .05.

^**

p < .01.

^***

p < .001.