Table 5. Post-hoc analysis: Structural equation model coefficients (weak invariance model).

		DV: Loneliness during lockdown
Category	Variable	ß		SE
Social media	Change in use of SM (T2 –T1)	.078	*	.044
usage	Baseline use of SM (T1)	.072	*	.039
Other interaction	Change in F2F communication (T2 –T0)	-.112	**	.016
channels	Baseline F2F communication (T1)	-.077	*	.019
	Change VC communication (T2 –T0)	-.003		.031
	Baseline VC communication (T1)	.012		.037
Other media	Change in use of other media (T2 –T1)	.017		.009
consumption	Baseline use of other media (T1)	.024		.009
Controls	Personal COVID-19 restraints (T2)	.213	**	.028
	Work-life COVID-19 restraints (T2)	-.018		.081
	Conscientiousness (T2)	-.034		.033
	Agreeableness (T2)	.060	*	.024
	Extraversion (T2)	.031		.027
	Neuroticism (T2)	.093	**	.026
	Openness (T2)	-.054	*	.025
	Age (T0)	-.076	*	.044
	Female (T0)	.017		.077
	Educational attainment (T0)	-.022		.057
	Unemployed (T2)	.014		.167
	Retired (T2)	.071	*	.120
	Living alone (T2)	.035		.089
	Social media confidence (T2)	-.022		.025
	Loneliness (pre-lockdown) (T1)	.616	**	.033
Interactions	Change in use of SM x	.003		.010
	Change in F2F communication
	Change in use of SM x	-.088	**	.018
	Change in VC communication

Note. R² = .600. ß = standardized coefficient; SE = standard error; SM = social media; F2F = face-to-face; VC = video chat. Number of observations: 825

*p < .05

** p < .01 (two tailed)