Table 3.

Communications reciprocity after earthquake.

Dependent variable = reciprocity	Coef.	Cluster SE	t	p > |t|
Ego and family all embedded	NA	NA	NA	NA
Earthquake dummy (1 = post-quake)	0.225	0.034	6.58	<0.001***
Ego and family all embedded X earthquake dummy	0.186	0.082	2.27	0.023*
ln(Degree centrality of ego)	−0.731	0.052	−14.06	<0.001***
ln(Total call frequency of ego)	1.293	0.035	36.97	<0.001***
ln(Total text frequency of ego)	0.133	0.025	5.32	<0.001***
ln(Internet usage frequency of ego)	0.015	0.006	2.69	0.007**
ln(Phone retail price in Yuan)	0.168	0.035	4.74	<0.001***
ln(Total WeChat frequency)	−0.005	0.018	−0.26	0.795
ln(Total frequency of other instant messaging)	−0.015	0.014	−1.03	0.303
Roaming dummy (1 = traveling outside of prefecture)	−0.054	0.068	−0.79	0.428
Rural dummy (1 = rural)	0.092	0.167	0.55	0.583
Damage dummy (1 = cell towers damaged in neighborhood)	−1.096	0.338	−3.25	0.001**
Prior intra-family outbound call share share	−0.076	0.170	−0.44	0.657
Family fixed effects	Yes

Notes: number of obs. = 49,851; time periods =  5; the p values are obtained from a two-sided t test; model controls for fixed effects of families; SE’s clustered at family levels; first variable is perfectly collinear with family fixed effects, and thus was excluded from model estimation.

Model 3 tested if fully embedded families had more reciprocal communications after the earthquake. We counted total number of reciprocal communications across the span of our panel data, where a call is reciprocal if the alter of an outbound (inbound) call was also the same alter of the next inbound (outbound) call (no distinction made between family and friends). The significant positive interaction term shows that the earthquake increased quantity of reciprocal communications for people with fully embedded family structures, relative to those with completely unembedded family structures.

^***p < 0.001, ^**p < 0.01, ^*p < 0.05.