. 2021 Jun 3;37(1):82–102. doi: 10.1177/02673231211012141

Table 1.

Bootstrapped OLS regression with three selective exposure measures.

	Self-reported selective exposure		Perceived selective exposure		Actual selective exposure
	B	SE B	B	SE B	B	SE B
(Constant)	2.479	.086	9.251	.295	.057	.074
Sex (baseline male)	.030	.021	.018^†	.010	−.012	.021
Age	.000	.002	.001	.002	.002^*	.001
Education	.033	.031	.097	.066	−.031^**	.010
Political interest	.033^*	.013	.057^*	.026	−.013^*	.007
Political orientation	.004	.005	−.075^†	.027	.015^***	.002
Political extremity	.083^***	.076	−.655^***	.042	.012^*	.005
Predictors #1 - #4
#1 Users in more fragmented-polarized media environments	.405^***	.059	.204^*	.099	−.047^*	.024
#2 TV user	−.092	.062	−.137	.095	−.066^**	.024
#3 Radio user	−.014	.023	−.013	.091	−.021^*	.082
#3 Newspaper user	−.042^†	.025	−.063	.097	.022	.027
#4 Social media user	.073^*	.029	−.02	.037	−.017	.011
N	2812		2560		2825
R ²	.101		.270		.018
F	299.48^***		804.78^***		51.04^***

OLS: ordinary least squares.

Estimates are unstandardized coefficients (B) with standard errors (SE B).

^***

p < .001, **p < .01, *p < .05, †p < .1 (two-tailed).