Table 1.

Parameter estimates and confidence intervals for the three data examples

Third Variable	Effect	Parameter	Estimate	Standard error	Est/SE	LCL	UCL
Mediation	Group on Recall	bYX	2.517	1.084	2.32	.329	4.705
	Group on Recall	bYX⋅T	.332	1.292	.26	-2.277	2.940
	Adjusted for imagery
	Imagery on Recall	bYT⋅X	.614	.226	2.72	.158	1.071
	Adjusted for group
	Group on imagery	bTX	3.558	.689	5.16	2.167	4.950
	Mediated effect	bTXbYT⋅X = bYX – bYX⋅T	2.185	.922	2.40	.631	3.847
Confounding	BMI on bench press	bYX	12.746	1.697	7.51	9.414	16.078
	BMI on bench press	bYT⋅T	10.916	1.649	6.62	7.678	14.153
	Adjusted for age
	Age on bench press	bYT⋅X	53.233	6.706	7.94	40.069	66.397
	Adjusted for BMI
	BMI on age	bTX	.034	.009	3.92	.017	.052
	Confounder effect	bTXbYT⋅X = bYX – bYX⋅T	1.830	.524	3.49	.854	2.933
Collision	Conscientiousness on Income	bYX	.022	.006	3.70	.010	.033
	Conscientiousness on	bYT⋅T	.017	.006	2.90	.006	.029
	Income adjusted for health
	Health on income	bYT⋅X	.041	.007	6.00	.028	.055
	Adjusted for conscientiousness
	Conscientiousness on	bTX	.112	.012	9.59	.089	.134
	health
	Collider effect	bTXbYT⋅X = bYX – bYX⋅T	.005	.001	5.00	.003	.007

Bootstrap samples (1000 samples) were used to compute confidence limits for mediated, confounder, and collider effects. Other confidence limits were based on the normal distribution computed with estimates and standard errors. Table entries are rounded estimates

Est estimate, SE standard error, LCL lower confidence limit, UCL upper confidence limit