Table 4.

Negative binomial regression on the number of stable, new, and lost core discussants (N = 2478).

	(1) # Stable core discussants		(2) # New core discussants		(3) # Lost core discussants
	IRR	SE	IRR	SE	IRR	SE
Social media use w1	1.036**	.011	1.022*	0.010	1.031***	0.011
Difference social media use w1, w2	1.023**	.009	1.025**	0.008	1.024**	0.009
# Stable core discussants			0.779***	0.015	0.743***	0.016
# New core discussants	0.849***	.013			1.039**	0.015
# Lost core discussants	0.847***	.013	1.029*	0.013
Female	1.187***	.044	1.092*	0.035	1.162***	0.042
Age	0.990**	.003	0.994*	0.002	0.996	0.003
Moroccan immigranta	0.560***	.037	0.888*	0.044	0.804***	0.045
Turkish immigranta	0.656***	.039	0.916	0.043	0.890*	0.046
Non-Western immigranta	0.791*	.083	0.973	0.084	0.952	0.091
Western immigranta	0.926	.075	0.988	0.073	0.939	0.079
Intermediate educationb	1.211***	.055	1.078**	0.042	1.193***	0.053
Higher educationb	1.387***	.070	1.172***	0.053	1.370***	0.069
Partner	1.081	.053	0.840***	0.035	0.799***	0.037
Time between t1 and t2	0.990	.036	1.019	0.032	1.086*	0.039
Web survey in w2	1.174***	.043	1.402***	0.046	0.847***	0.016
Intercept	1.956***	.331	1.868*	0.281	1.376	0.235

IRR: incidence rate ratio; SE: standard error.

^*p < .05; **p < .01; ***p < .001.

^aRef.: Native Dutch.

^bRef.: Lower education.