Peer effects of depression between left-behind and non-left-behind children: quasi-experimental evidence from rural China

Table 3 Results of sub-sample estimates

Variables	LB children		NLB children
	(1)	(2)	(3)	(4)
	Extraversion	Non-extraversion	Extraversion	Non-extraversion
Panel A: Teacher personality
LB classmates’ depression scores	− 5.526***	− 0.958*	− 9.835**	0.442***
LB classmates’ depression scores	(1.146)	(0.380)	(2.629)	(0.112)
NLB classmates’ depression scores	− 9.438***	1.310***	− 14.286**	− 0.165
NLB classmates’ depression scores	(1.789)	(0.346)	(3.997)	(0.235)
Observations	100	383	236	1,098
R-squared	0.808	0.306	0.741	0.232

	LB children		NLB children
	(1)	(2)	(3)	(4)
	Active	Not active	Active	Not active
Panel B: Active parent–child communication
LB classmates’ depression scores	− 0.166	− 0.657*	0.430*	0.339**
LB classmates’ depression scores	(0.228)	(0.299)	(0.207)	(0.122)
NLB classmates’ depression scores	0.858***	0.719**	0.207*	0.354**
NLB classmates’ depression scores	(0.173)	(0.223)	(0.082)	(0.103)
Observations	223	260	636	698
R-squared	0.314	0.315	0.193	0.223

	LB children		NLB children
	(1)	(2)	(3)	(4)
	High quantity	Low quantity	High quantity	Low quantity
Panel C: Friendship
LB classmates’ depression scores	− 0.180	− 0.802	0.143*	0.518***
LB classmates’ depression scores	(0.196)	(0.433)	(0.059)	(0.141)
NLB classmates’ depression scores	0.623***	1.165**	0.534***	0.133
NLB classmates’ depression scores	(0.144)	(0.403)	(0.148)	(0.150)
Observations	276	207	784	550
R-squared	0.227	0.329	0.182	0.199

	LB children		NLB children
	(1)	(2)	(3)	(4)
	High quality	Low quality	High quality	Low quality
LB classmates’ depression scores	− 0.100	− 0.402	0.081	0.486**
LB classmates’ depression scores	(0.270)	(0.282)	(0.071)	(0.170)
NLB classmates’ depression scores	0.459*	0.814**	0.490	0.110
NLB classmates’ depression scores	(0.218)	(0.245)	(0.270)	(0.153)
Observations	224	259	687	647
R-squared	0.237	0.322	0.158	0.219
Student and parent controls	Yes	Yes	Yes	Yes
Correlated effects controls	Yes	Yes	Yes	Yes
School-grade fixed effects	Yes	Yes	Yes	Yes

Student and parent controls include the child’s gender, age, only-child status, boarding-student status, adjusted self-rated health, mental development, time spent playing online games, guardian’s education, parent’s occupation and family economic status. Correlated effect controls include class size, head teacher’s career duration, gender, education, professional job title and age. Standard errors are clustered at the class level and reported in parentheses. The weight of the inclusion is 1/sample. Significance: *** p < 0.001, ** p < 0.01, * p < 0.05

ISSN: 1753-2000