Teamwork and Human Capital Development

Chunchao Wang; Aiping Xiao; Yu Zhou

doi:10.3368/jhr.0121-11400R2

Article Figures & Data

Figures

Tables
Additional Files

Download figure
Open in new tab
Download powerpoint
Figure 1
Seating Arrangement in a Typical Classroom
Download figure
Open in new tab
Download powerpoint
Figure 2
Experimental Procedure

Tables

Figures
Additional Files

View popup

Table 1

Gender Composition of Sample Schools

		Number		Percentage
School Code	Total no. of Students	Male	Female	Male	Female	Treatment Class Code	Control Class Code
HN01	258	136	122	53%	47%	III (2)	III (1)
						IV (2)	IV (1)
						V (2)	V (1)
HN02	304	167	137	55%	45%	III (2)	III (1)
						IV (2)	IV (1)
						V (2)	V (1)
HN04	338	187	151	55%	45%	III (2)	III (4)
						IV (2)	IV (1)
						V (1)	V (2)
HN05	365	218	147	60%	40%	III (2)	III (1)
						IV (2)	IV (1)
						V (2)	V (1)
HN06	324	199	125	61%	39%	III (117)	III (118)
						IV (113)	IV (115)
						V (111)	V (112)
Total	1,589	907	682	57%	43%

Data source: Experimental data.
Notes: In last two columns, Roman numerals denote grade years, and Arabic numbers enclosed in parentheses refer to class codes.

View popup

Table 2

Descriptive Statistics of Treatment and Control Classes

	Treatment Classes			Control Classes
Variables	Mean	SD	N	Mean	SD	N	Mean Differences
Average baseline score	55.112	18.418	790	55.260	19.308	799	−0.148
Noncognitive Skills
Openness	0.015	1.018	775	−0.015	0.982	783	0.030
Conscientiousness	0.011	1.026	775	−0.012	0.977	783	0.023
Extraversion	−0.011	1.018	775	0.013	0.985	783	−0.024
Agreeableness	−0.049	1.040	775	0.049	0.957	783	−0.098^*
Neuroticism	0.074	1.011	775	−0.077	0.035	783	0.151^***
Individual Characteristics
Gender (1 = male, 0 = female)	0.568	0.496	790	0.573	0.495	799	−0.005
Age (year)	9.501	1.070	790	9.503	1.009	799	−0.002
Height (cm)	135.306	11.621	790	134.585	9.463	799	0.722

Data source: Experimental data.
Notes: Noncognitive skills are standardized to have a mean of zero and SD of one. Significance: ^*p < 0.1, ^***p < 0.01.

View popup

Table 3

Descriptive Statistics of Variables

Variables	Mean	SD	Min.	Max.	N
Cognitive Skills
Standardized baseline scores	0	1	−2.93	2.08	1,589
Standardized midterm scores	0	1	−3.02	2.33	1,589
Standardized ﬁnal scores	0	1	−3.01	2.27	1,589
Noncognitive Skills
Openness	0	1	−4.95	5.26	3,043
Conscientiousness	0	1	−3.61	4.63	3,043
Extraversion	0	1	−4.03	5.27	3,043
Agreeableness	0	1	−2.66	4.29	3,043
Neuroticism	0	1	−4.90	3.36	3,043
Control Variables
Treatment × Post	0.25	0.43	0	1	3,178
Treatment team dummy	0.50	0.50	0	1	3,178
Extracurricular reading time	11.74	17.56	0	300	3,036
Household chore time	12.36	20.51	0	288	3,035
Parents’ yearly income	1.89	0.59	0	5.19	3,011

Data source: Experimental data.
Notes: Extracurricular reading time and household chore time are measured in minutes per day. Parents’ yearly income is the logarithm of RMB 10,000.

View popup

Table 4

Effects of Forming Teams on Cognitive Skills

	Midterm Examination				Final Examination
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
Panel A: OLS Estimate
Treatment	0.073^***	0.077^***	0.078^***	0.095^***	0.072^***	0.075^***	0.076^***	0.082^***
	(0.013)	(0.013)	(0.014)	(0.018)	(0.018)	(0.020)	(0.021)	(0.025)
Baseline		0.816^***	0.816^***	0.810^***		0.791^***	0.791^***	0.788^***
		(0.026)	(0.026)	(0.027)		(0.024)	(0.024)	(0.024)
Individual characteristics		Yes	Yes	Yes		Yes	Yes	Yes
Class characteristics			Yes	Yes			Yes	Yes
Family characteristics				Yes				Yes
N	1,589	1,589	1,589	1,462	1,589	1,589	1,589	1,462
Adjusted R²	0.004	0.700	0.699	0.695	0.001	0.685	0.683	0.686
Panel B: DID Estimate
Treatment	−0.018	−0.012	−0.011	−0.011	−0.017	−0.012	−0.009	−0.009
	(0.014)	(0.012)	(0.012)	(0.014)	(0.014)	(0.012)	(0.011)	(0.012)
Post	−0.023^*	−0.023^*	−0.024	−0.026	−0.033	−0.033	−0.032	−0.028
	(0.014)	(0.012)	(0.015)	(0.015)	(0.020)	(0.020)	(0.021)	(0.021)
Treatment × Post	0.090^***	0.090^***	0.091^***	0.100^***	0.089^***	0.089^***	0.087^***	0.085^***
	(0.015)	(0.015)	(0.015)	(0.018)	(0.023)	(0.023)	(0.023)	(0.025)
Baseline		0.909^***	0.909^***	0.908^***		0.897^***	0.897^***	0.898^***
		(0.012)	(0.012)	(0.013)		(0.012)	(0.0123)	(0.012)
Individual characteristics		Yes	Yes	Yes		Yes	Yes	Yes
Class characteristics			Yes	Yes			Yes	Yes
Family characteristics				Yes				Yes
N	3,178	3,178	3,178	3,003	3,178	3,178	3,178	3,003
Adjusted R²	0.001	0.829	0.828	0.828	0.001	0.820	0.820	0.823
Finite sample inference p				0.000				0.000

Notes: Each column is a separate regression. Panel A reports the OLS estimates using test scores in midterm examination (Columns 1–4) and final examination (Columns 5–8) as dependent variables. Panel B reports the results of DID estimates from Equation 1 using midterm examination scores in Columns 1–4 and final examination scores in Columns 5–8. Individual characteristics are gender, age, and height. Class characteristics are school, grade, class, and team dummy variables. Family characteristics are parents’ income, extracurricular reading time, and household chore time. Standard errors are clustered at the school–grade level and enclosed in parentheses. The last row reports finite sample inference. Significance: ^*p < 0.1, ^***p < 0.01.

View popup

Table 5

Heterogeneous Effects of Forming Teams: Cognitive Skills

	Gender		Academic Performance				Grade
	Male	Female	A	B	C	D	Third	Fourth	Fifth
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
Midterm	0.128^***	0.065^*	0.240	0.046	0.043	0.086^**	0.186^***	0.116^*	0.037^**
	(0.035)	(0.035)	(0.184)	(0.052)	(0.364)	(0.041)	(0.019)	(0.038)	(0.011)
Between group test	p(female = male) = 0.3001		p(A = B) = 0.4957 p(A = C) = 0.5727				p(third = fourth) = 0.3666
			p(A = D) = 0.5377 p(B = C) = 0.9694				p(third = fifth) = 0.0322
			p(B = D) = 0.4338 p(C = D) = 0.4448				p(fourth = fifth) = 0.3107
Final	0.057	0.105^*	0.276	0.068	0.047	0.063^**	0.165^***	0.119	0.003
	(0.043)	(0.049)	(0.236)	(0.082)	(0.068)	(0.031)	(0.024)	(0.059)	(0.020)
Between group test	p(female = male) = 0.4294		p(A = B) = 0.8110 p(A = C) = 0.9432				p(third = fourth) = 0.5594
			p(A = D) = 0.3412 p(B = C) = 0.7995				p(third = fifth) = 0.0221
			p(B = D) = 0.3135 p(C = D) = 0.3654				p(fourth = fifth) = 0.1322

Notes: Each column reports the results of a cohort with the specification including individual characteristics, school characteristics, and family characteristics controlled. Midterm represents the DID analysis comparing midterm examination scores with baseline scores. Final represents the DID analysis comparing final examination scores with baseline scores. A, B, C, and D denote levels of academic performance from high to low. Standard errors are clustered at the school–grade level and enclosed in parentheses. Between‐group test reports the p‐value from the between‐group t‐test. Significance: ^*p < 0.1, ^**p < 0.05, ^***p < 0.01.

View popup

Table 6

Between‐Group Comparison of Noncognitive Skills

	First‐Round Questionnaire			Second‐Round Questionnaire
	Treatment Classes	Control Classes	Mean Differences	Treatment Classes	Control Classes	Mean Differences
Openness	0.015	−0.015	0.030	0.034	−0.035	0.070
Conscientiousness	0.011	−0.011	0.023	0.112	−0.107	0.219^***
Extraversion	−0.011	0.013	−0.024	0.066	−0.063	0.129^**
Agreeableness	−0.049	0.049	−0.098^*	−0.073	0.068	−0.140^***
Neuroticism	0.074	−0.077	0.151^***	0.115	−0.108	0.223^***

Data source: Experimental data.
Notes: Significance: ^*p < 0.1, ^**p < 0.05, ^***p < 0.01.

View popup

Table 7

Effects of Forming Teams on Noncognitive Skills

	OLS Estimates				DID Estimates
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
Agreeableness	−0.138^*	−0.127^*	−0.147^***	−0.141^**	−0.116^*	−0.122^*	−0.133^**	−0.149^**	0.020
	(0.075)	(0.073)	(0.055)	(0.054)	(0.063)	(0.063)	(0.062)	(0.063)
Conscientiousness	0.219^***	0.215^**	0.222^**	0.217^**	0.209^**	0.207^***	0.154^**	0.140^*	0.000
	(0.052)	(0.105)	(0.105)	(0.098)	(0.087)	(0.072)	(0.070)	(0.078)
Extraversion	0.129^**	0.118	0.123	0.115	0.156	0.147	0.140	0.130	0.100
	(0.052)	(0.099)	(0.111)	(0.122)	(0.109)	(0.122)	(0.116)	(0.109)
Neuroticism	0.223^***	0.214^***	0.229^***	0.226^***	0.170^***	0.172^***	0.162^**	0.200^***	0.010
	(0.076)	(0.075)	(0.062)	(0.063)	(0.065)	(0.065)	(0.065)	(0.065)
Openness	0.070	0.072	0.060	0.065	0.041	0.046	0.056	0.057	0.190
	(0.052)	(0.089)	(0.087)	(0.091)	(0.030)	(0.036)	(0.041)	(0.035)
Individual characteristics		Yes	Yes	Yes		Yes	Yes	Yes
School characteristics			Yes	Yes			Yes	Yes
Family characteristics				Yes				Yes

Notes: Each column is a separate regression. Columns 1–4 use OLS specifications using only the personality traits from the second‐round questionnaire, and Columns 5–8 use DID specifications. Column 9 reports the finite sample inference p‐value of the full specification in Column 8. Individual characteristics are gender, age, and height. Class characteristics are school, grade, class, and team dummy variables. Family characteristics are parents’ income, extracurricular reading time, and household chore time. Standard errors are clustered at the school–grade level and enclosed in parentheses. Significance: ^*p < 0.1, ^**p < 0.05, ^***p < 0.01.

View popup

Table 8

Heterogeneous Effects of Forming Teams: Noncognitive Skills

	Gender		Academic Performance				Grade
	Male	Female	A	B	C	D	Third	Fourth	Fifth
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
Openness	0.105	0.003	0.092	−0.045	0.149	−0.315	0.104	0.038	0.125
Openness	(0.146)	(0.125)	(0.101)	(0.215)	(0.291)	(0.500)	(0.150)	(0.167)	(0.170)
Between group test	p(female = male) = 0.1306		p(A = B) = 0.3093 p(A = C) = 0.9650				p(third = fourth) = 0.9963
			p(A = D) = 0.7003 p(B = C) = 0.5421				p(third = fifth) = 0.4476
			p(B = D) = 0.8993 p(C = D) = 0.7635				p(fourth = fifth) = 0.4671
Conscientiousness	0.174	0.116	−0.266	−0.134	0.179	0.199^**	0.297^**	0.303^*	0.023
Conscientiousness	(0.112)	(0.131)	(0.825)	(0.271)	(0.158)	(0.088)	(0.145)	(0.176)	(0.151)
Between group test	p(female = male) = 0.7086		p(A = B) = 0.4304 p(A = C) = 0.1079				p(third = fourth) = 0.6363
			p(A = D) = 0.2497 p(B = C) = 0.3931				p(third = fifth) = 0.4461
			p(B = D) = 0.4318 p(C = D) = 0.7902				p(fourth = fifth) = 0.2268
Extraversion	0.208^*	0.050	0.303	0.195	0.029	0.149	0.297^**	0.303^*	0.023
Extraversion	(0.117)	(0.129)	(0.831)	(0.243)	(0.158)	(0.126)	(0.145)	(0.176)	(0.151)
Between group test	p(female = male) = 0.3260		p(A = B) = 0.5783 p(A = C) = 0.7644				p(third = fourth) = 0.9613
			p(A = D) = 0.2254 p(B = C) = 0.5289				p(third = fifth) = 0.4389
			p(B = D) = 0.1698 p(C = D) = 0.3382				p(fourth = fifth) = 0.5097
Agreeableness	−0.092	−0.183^*	−0.156	−0.229	−0.222^*	−0.113	0.009	−0.022	−0.271^**
Agreeableness	(0.088)	(0.101)	(0.779)	(0.224)	(0.125)	(0.079)	(0.173)	(0.151)	(0.127)
Between group test	p(female = male) = 0.2323		p(A = B) = 0.3513 p(A = C) = 0.3222				p(third = fourth) = 0.6427
			p(A = D) = 0.9228 p(B = C) = 0.8259				p(third = fifth) = 0.0923
			p(B = D) = 0.6562 p(C = D) = 0.5861				p(fourth = fifth) = 0.0454
Neuroticism	0.122	0.231^**	−0.095	0.396^*	0.201	0.151^*	0.169	0.092	0.224^*
Neuroticism	(0.090)	(0.105)	(0.600)	(0.218)	(0.138)	(0.078)	(0.119)	(0.209)	(0.125)
Between group test	p(female = male) = 0.2596		p(A = B) = 0.2765 p(A = C) = 0.2739				p(third = fourth) = 0.1142
			p(A = D) = 0.6667 p(B = C) = 0.8286				p(third = fifth) = 0.6508
			p(B = D) = 0.4035 p(C = D) = 0.3590				p(fourth = fifth) = 0.0584

Notes: Each column reports the results of a cohort with the specification including individual characteristics, school characteristics, and family characteristics controlled. A, B, C, and D denote levels of academic performance from high to low. Standard errors are clustered at the school–grade level and enclosed in parentheses. The between‐group test reports the p‐value from the between‐group t‐test. Significance: ^*p < 0.1, ^***p < 0.01.

View popup

Table 9

Between‐Group Comparison of Class Behaviors

	First Round			Second Round
In‐Class Discipline and Attention	Treatment Classes	Control Classes	Mean Differences	Treatment Classes	Control Classes	Mean Differences
Unpermitted talking	3.225	3.867	−0.642^**	3.392	6.857	−3.465^***
Pranking	3.094	3.909	−0.815^***	3.165	6.304	−3.139^***
Reading comics	0.741	0.170	0.571^***	0.997	1.556	−0.558^*
Sleeping	0.250	0.088	0.162^***	1.125	1.472	−0.347
Tardiness	0.106	0.062	0.043^**	1.108	1.497	−0.388
In‐class attention	1.547	1.489	0.059	1.693	1.503	0.190^***

Notes: The measurement for talking without permission, pranking, reading comic books in class, sleeping in class, and tardiness is frequency per week. In‐class attention is measured on three levels; that is, 0 = mostly cannot pay attention, 1 = not sure, or 2 = mostly can pay attention. Significance: ^*p < 0.1, ^**p < 0.05, ^***p < 0.01.

View popup

Table 10

Effects of Forming Teams on In‐Class Discipline and Class Attention

	Talk	Prank	Comics	Sleep	Tardiness	Attention
	(1)	(2)	(3)	(4)	(5)	(6)
Treatment × Post	−1.961^***	−1.563^***	−0.801^***	0.000	−0.006	0.234^***
	(0.293)	(0.287)	(0.303)	(0.013)	(0.016)	(0.047)
Treatment	0.218	−0.172	0.274^**	0.016^*	0.028^**	−0.021
	(0.228)	(0.234)	(0.130)	(0.009)	(0.011)	(0.039)
Post	1.189^***	0.748^***	1.106^***	0.018^**	0.024^***	−0.023
	(0.200)	(0.194)	(0.212)	(0.008)	(0.009)	(0.035)
Individual characteristics	Yes	Yes	Yes	Yes	Yes	Yes
Class characteristics	Yes	Yes	Yes	Yes	Yes	Yes
Family characteristics	Yes	Yes	Yes	Yes	Yes	Yes
N	2,580	2,580	2,669	2,469	2,489	2,049
Adjusted R²	0.153	0.170	0.053	0.017	0.021	0.200

Notes: Each column reports the results of a separate equation. The measurement for talking without permission, pranking, reading comic books in class, sleeping in class, and tardiness is times per week. In‐class attention is measured on three levels: 0 = mostly cannot pay attention, 1 = not sure, and 2 = mostly can pay attention. Individual characteristics are gender, age, and height. Class characteristics are school, grade, class, and team dummy variables. Family characteristics are parents’ income, extracurricular reading time, and household chore time. Standard errors are clustered at the school–grade level and enclosed in parentheses. Significance: ^*p < 0.1, ^**p < 0.05, ^***p < 0.01.

View popup

Table 11

Descriptive Statistics of Treatment Classes and Survey Sample

	Treatment Classes			Random Sample			Mean Differences
Variables	Mean	SD	N	Mean	SD	N	Mean Differences
Cognitive Skills
Standardized baseline scores	−0.001	0.991	791	−0.004	1.005	73	0.003
Noncognitive Skills
Openness	0.015	1.019	728	−0.031	0.881	73	0.065
Conscientiousness	0.112	1.054	728	0.062	0.795	73	0.051
Extraversion	0.067	1.074	728	0.085	0.957	73	−0.018
Agreeableness	−0.074	1.087	728	−0.174	0.933	73	0.100
Neuroticism	0.116	1.047	728	0.094	0.868	73	0.022
Individual Characteristics
Gender	0.568	0.496	790	0.589	0.495	73	−0.021
Age	9.501	1.070	790	9.521	1.042	73	−0.020
Height	135.306	11.621	790	135.658	8.39	73	−0.352
Grade	4.006	0.821	790	3.986	0.842	73	0.020
School	3.061	1.448	790	3.123	1.443	73	−0.063

Data source: Experiment data.
Notes: Cognitive skills and noncognitive skills are standardized to have a mean of zero and SD of one.

View popup

Table 12

Follow‐up Survey on Class Teachers

Question	Strongly Agree	Agree	Neither Agree nor Disagree	Disagree	Total
Connection	9 (60%)	5 (33%)	1 (7%)	0 (0%)	15
Mutual help	6 (40%)	8 (53%)	1 (7%)	0 (0%)	15
Study supervision	8 (53%)	5 (33%)	2 (13%)	0 (0%)	15
Discipline supervision	3 (20%)	10 (67%)	2 (13%)	0 (0%)	15
Competition	6 (40%)	8 (53%)	1 (7%)	0 (0%)	15
Test score	7 (47%)	5 (33%)	2 (13%)	1 (7%)	15
In‐class discipline	2 (13%)	8 (53%)	4 (27%)	1 (7%)	15
Attention	0 (0%)	2 (13%)	5 (33%)	8 (53%)	15
Curriculum	0 (0%)	1 (7%)	4 (27%)	10 (67%)	15
Pedagogy	0 (0%)	0 (0%)	5 (33%)	10 (67%)	15
Other class	0 (0%)	0 (0%)	7 (47%)	8 (53%)	15

Data source: Survey data.
Notes: For each entry, the number is the count of the corresponding answer. Percentage in parentheses is the count of corresponding answer divided by total count of all answers.

View popup

Table 13

Follow‐up Survey on Students

Question	Strongly Agree	Agree	Neither Agree nor Disagree	Disagree	Strongly Disagree	Total
Adapt time	22 (30%)	36 (49%)	13 (18%)	2 (3%)	0 (0%)	73
Make friends	22 (30%)	40 (55%)	7 (10%)	4 (5%)	0 (0%)	73
Team continue	19 (26%)	36 (49%)	15 (21%)	3 (4%)	0 (0%)	73
Responsible	36 (49%)	32 (44%)	4 (5%)	1 (1%)	0 (0%)	73
Think action	28 (38%)	40 (55%)	5 (7%)	0 (0%)	0 (0%)	73
More effort	33 (45%)	33 (45%)	7 (10%)	0 (0%)	0 (0%)	73
Friend easily	31 (42%)	33 (45%)	9 (12%)	0 (0%)	0 (0%)	73
After school	28 (38%)	31 (42%)	12 (16%)	1 (1%)	1 (1%)	73
Study self‐policing	32 (44%)	34 (47%)	7 (10%)	0 (0%)	0 (0%)	73
Discipline self‐policing	35 (48%)	33 (45%)	4 (5%)	1 (1%)	0 (0%)	73
Study from stigma	23 (32%)	32 (44%)	15 (21%)	3 (4%)	0 (0%)	73
Discipline from stigma	28 (38%)	28 (38%)	15 (21%)	2 (3%)	0 (0%)	73
Study role model	33 (45%)	29 (40%)	10 (14%)	1 (1%)	0 (0%)	73
Discipline role model	33 (45%)	30 (41%)	9 (12%)	1 (1%)	0 (0%)	73
Cooperation	35 (48%)	32 (44%)	6 (8%)	0 (0%)	0 (0%)	73
Kolmogorov–Smirnov test on reasons for studying
p(self policing = social stigma) = 0.379
p(self policing = role model) = 1.000
p(social stigma = role model) = 0.500
Kolmogorov–Smirnov test on reasons for observing in‐class discipline
p(self policing = social stigma) = 0.277
p(self policing = role model) = 0.996
p(social stigma = role model) = 0.890

Data source: Survey data.
Notes: For each entry, the number is the count of the corresponding answer. Percentage in parentheses is the count of the corresponding answer divided by the total count of all answers.

View popup

Table 14

Correlation between Baseline Test Scores and Noncognitive Skills

	(1)	(2)	(3)	(4)	(5)	(6)
Openness	0.118^***					0.048^**
	(0.018)					(0.017)
Conscientiousness		0.211^***				0.156^***
		(0.030)				(0.034)
Extraversion			0.136^***			0.037
			(0.026)			(0.032)
Agreeableness				0.109^**		0.039
				(0.039)		(0.031)
Neuroticism					−0.143^***	−0.082^***
					(0.024)	(0.021)
Individual characteristics	Yes	Yes	Yes	Yes	Yes	Yes
Class characteristics	Yes	Yes	Yes	Yes	Yes	Yes
Family characteristics	Yes	Yes	Yes	Yes	Yes	Yes
N	1,540	1,540	1,540	1,540	1,540	1,540
Adjusted R²	0.055	0.083	0.059	0.051	0.059	0.094

Notes: Each column is a separate regression. Individual characteristics are gender, age, and height. Class characteristics are school, grade, class, and team dummy variables. Family characteristics are parents’ income, extracurricular reading time, and household chore time. Standard errors are clustered at the school–grade level and enclosed in parentheses. Significance: ^**p < 0.05, ^***p < 0.01.

View popup

Table 15

Treatment Effect with Controls on Noncognitive Skills

	(1)	(2)	(3)	(4)
Treatment	−0.009	−0.001	0.001	0.000
	(0.036)	(0.003)	(0.004)	(0.006)
Post	−0.020	0.000	0.001	−0.001
	(0.037)	(0.005)	(0.005)	(0.004)
Treatment × Post	0.061^*	0.079^***	0.080^***	0.083^***
	(0.034)	(0.018)	(0.019)	(0.019)
Openness	0.068^***	0.012^*	0.015^**	0.016^**
	(0.018)	(0.006)	(0.006)	(0.006)
Conscientiousness	0.116^***	0.005	0.004	0.004
	(0.013)	(0.007)	(0.008)	(0.008)
Extraversion	0.026	−0.002	−0.002	−0.001
	(0.023)	(0.007)	(0.007)	(0.007)
Agreeableness	−0.004	0.001	0.006	0.005
	(0.025)	(0.007)	(0.007)	(0.007)
Neuroticism	−0.075^***	0.001	0.000	0.000
	(0.025)	(0.008)	(0.009)	(0.009)
Individual characteristics		Yes	Yes	Yes
Class characteristics			Yes	Yes
Family characteristics				Yes
N	3,045	2,988	2,988	2,980
Adjusted R²	0.039	0.850	0.850	0.850

Notes: Each column is a separate regression. Individual characteristics are gender, age, and height. Class characteristics are school, grade, class, and team dummy variables. Family characteristics are parents’ income, extracurricular reading time, and household chore time. Standard errors are clustered at the school–grade level and enclosed in parentheses. Significance: ^*p < 0.1, ^**p < 0.05, ^***p < 0.01.

View popup

Table A1

Runs Test on Survey Sample

Variables	Test Value	Cases < Test Value	Cases ≥ Test Value	Total Cases	Number of Runs	z‐Score	p‐Value
Baseline scores	−0.004	33	38	71	30	−1.520	0.129
Noncognitive Skills
Openness	−0.031	37	35	72	42	1.194	0.232
Conscientiousness	0.062	32	40	72	42	1.309	0.191
Extraversion	0.085	36	36	72	36	−0.237	0.812
Agreeableness	−0.174	31	41	72	35	−0.316	0.752
Neuroticism	0.094	34	38	72	34	−0.688	0.492
Individual Characteristics
Gender	0.589	30	43	73	29	−1.788	0.074
Age	9.521	36	37	73	31	−1.531	0.126
Height	135.658	36	37	73	38	0.119	0.905
Grade	3.986	26	47	73	28	−1.667	0.095
School	3.123	43	30	73	30	−1.545	0.122

Data source: Experiment data.
Notes: Baseline scores and noncognitive skills are standardized to have a mean of zero and SD of one. Gender is a dummy variable that is one for male and zero for female. Height is measured in centimeters. School is a category variable with values one to five.