The Impact of Early Marriage on the Life Satisfaction, Education and Subjective Health of Young Women in India: A Longitudinal Analysis

Figures & data

Table 1. Age at marriage

Age married	Number of girls/women married at each age
11	3
12	3
13	11
14	14
15	34
16	43
17	31
18	37
19	39
20	40
21	18
22	1
Married by age 22	274 (56.38%)
Married before age 18	139 (28.6%)
Not married at age 22	212 (43.62%)
Total	486

Table 2. Summary statistics

	(1) All girls (n = 486)		(2) Girls married before age 18 (n = 139)		(3) Girls not married before age 18 (n = 347)
	Mean	SD	Mean	SD	Mean	SD
Life satisfaction, educational attainment and health
Life satisfaction age 12^a	3.63	1.67	3.08***	1.41	3.85	1.71
Child’s educational attainment age 8 (highest grade: 0–4)	1.87	0.89	1.88	0.84	1.87	0.90
Child’s educational attainment age 12 (highest grade: 0–9)	5.61	1.28	5.40**	1.35	5.70	1.24
Child’s subjective health age 8	1.39	0.66	1.43	0.65	1.38	0.66
Child’s subjective health age 12	1.36	0.65	1.42*	0.62	1.33	0.66
Child and household characteristics (round 1^b)
Age in months	95.9	3.76	96.09	3.71	95.82	3.79
Location: Urban	0.23	0.42	0.09***	0.29	0.29	0.45
Location: Rural	0.77	0.42	0.91***	0.29	0.71	0.45
Wealth index	0.40	0.21	0.35***	0.16	0.43	0.22
Household head: Female	0.08	0.28	0.06	0.23	0.09	0.29
Household Head: Male	0.92	0.28	0.94	0.23	0.91	0.29
Household Head Education: None	0.60	0.49	0.78***	0.41	0.53	0.50
Household Head Education: Basic	0.22	0.41	0.14***	0.35	0.25	0.44
Household Head Education: Secondary or above	0.18	0.38	0.08***	0.27	0.21	0.41
Household size	5.6	1.95	5.34*	1.54	5.64	2.08
Eldest older sibling is brother	0.34	0.47	0.45***	0.5	0.29	0.46
Eldest older sibling is sister	0.31	0.46	0.25*	0.44	0.33	0.47
Family shock (previous 4 years): Yes	0.31	0.46	0.34	0.48	0.29	0.46
Child’s religion: Hindu	0.88	0.33	0.91*	0.29	0.86	0.35
Child’s religion: Other	0.13	0.33	0.09*	0.29	0.14	0.35
Child’s caste: SC (scheduled)	0.21	0.42	0.22	0.42	0.20	0.40
Child’s caste: ST (scheduled tribes)	0.13	0.33	0.14	0.34	0.12	0.33
Child’s caste: BC (economically and socially backward classes)	0.44	0.50	0.52**	0.50	0.41	0.49
Child’s caste: OC (other/open category)	0.22	0.42	0.12***	0.33	0.27	0.44
Child’s region: Coastal Andhra Pradesh	0.34	0.48	0.26***	0.44	0.38	0.49
Child’s region: Rayalaseema	0.30	0.46	0.37**	0.49	0.27	0.45
Child’s region: Telangana	0.36	0.48	0.37	0.48	0.35	0.48
Height for age standardised z-score	−1.58	0.99	−1.42**	1.04	−1.64	0.97
BMI for age standardised z-score	−1.29	0.93	−1.29	−1.45	−1.30	0.91

Notes: ^aLife satisfaction unavailable round 1, first recorded in round 2 (age 12).

^b All reported characteristic means are for round 1 at age 8.

Significance in difference in means t tests for girls who did/did not marry before age 18: ***p < 0.01; **p < 0.05; *p < 0.10.

Table 3. Antecedents of early marriage

	(i)	(ii)
Dependent variable Independent variables^a	ChildMarriage (logit)	AgeMarriage (tobit)
Life satisfaction	−0.042***	0.639***
	(0.014)	(0.157)
Education (highest grade attained)	0.003	−0.223
	(0.024)	(0.288)
Subjective Health	−0.015	−0.060
	(0.031)	(0.389)
Age(months)	0.003	−0.023
	(0.005)	(0.066)
Location: Rural	0.141**	−1.423*
	(0.064)	(0.847)
Wealth Index	−0.026	1.310
	(0.149)	(1.756)
Household head (HH): Male	0.083	−2.882***
	(0.066)	(1.077)
HH Education: Basic	−0.103*	1.033
	(0.053)	(0.637)
HH Education: >Secondary	−0.152***	3.379***
	(0.058)	(0.861)
Household size	−0.013	0.269**
	(0.010)	(0.129)
Eldest Older Sibling: brother	0.108**	−1.487***
	(0.044)	(0.554)
Eldest Older Sibling: sister	0.016	−0.265
	(0.051)	(0.629)
Family shock	0.006	−0.436
	(0.044)	(0.553)
Child’s religion: Hindu	−0.000	0.121
	(0.064)	(0.835)
Child’s caste: ST	0.031	−0.913
	(0.075)	(0.954)
Child’s caste: BC	0.062	−1.799***
	(0.051)	(0.674)
Child’s caste: OC	−0.046	−0.153
	(0.061)	(0.854)
Child’s region: Rayalaseema	0.184***	−2.750***
	(0.055)	(0.700)
Child’s region: Telangana	0.083*	−2.062***
	(0.050)	(0.662)
Height for age standardised z-score	0.073***	−0.862***
	(0.020)	(0.233)
BMI for age standardised z-score	0.023	−0.449*
	(0.022)	(0.258)
Constant	–	24.183***
	–	(6.489)
Observations	474	488
(Pseudo) R-squared	0.157	0.0676
Χ^2, F	76.17***	7.79***

Notes: Figures are average marginal effects in estimation (i); coefficients in estimations (ii). Robust standard errors in parentheses. ***p < 0.01, **p < 0.05, *p < 0.1. ^aAll independent variables except life satisfaction recorded at age 8 (life satisfaction first recorded at age 12).

Table 4. Difference-in-differences – matched female sample

Panel (a): Life satisfaction (N = 542)
	Before	After	Total
Control	133	135	268
Treated	135	131	266
Total	268	266
Outcome var.	Life satisfaction	S. Err.	T	P > t
Pre-treatment (age 12)
Control	3.722
Treated	3.067
Diff (T-C)	−0.655	0.177	−3.69	0.00***
Post-treatment (age 22)
Control	5.015
Treated	4.679
Diff (T-C)	−0.335	0.178	−1.88	0.060*
Difference-in-Differences	0.320	0.251	1.27	0.204
R^2	0.22
Panel (b): educational attainment (N = 550)
	Before	After	Total
Control	137	136	273
Treated	135	134	269
Total	272	270
Outcome var.	Educational attainment	S. Err.	T	P > t
Pre-treatment (age 8)
Control	1.869
Treated	1.881
Diff (T-C)	0.013	0.281	0.05	0.964
Post-treatment (age 22)
Control	11.669
Treated	8.373
Diff (T-C)	−3.296	0.282	−11.67	0.000***
Difference-in-Differences	−3.309	0.399	−8.30	0.000***
R^2	0.77

Panel (c): Subjective health (N = 548)
	Before	After	Total
Control	137	135	272
Treated	137	131	268
Total	274	266
Outcome var.	Subjective health	S. Err.	T	P > t
Pre-treatment (age 8)
Control	1.372
Treated	1.431
Diff (T-C)	0.058	0.078	0.75	0.454
Post-treatment (age 22)
Control	1.622
Treated	1.420
Diff (T-C)	−0.202	0.079	−2.56	0.011**
Difference-in-Differences	−0.261	0.111	−2.35	0.019**
R^2	0.02

Note: *** p < 0.01; ** p < 0.05; * p < 0.1.

Figure 1. Relationship between life satisfaction, education, health, and age at marriage.

Figure A1. Matching process.

Figure A2. Relationship between education, health, and age at marriage.

Table A1. Variable definitions

Variable	Definition
ChildMarriage	Married before age 18:1 yes, 0 No. Dependent variable in Logit.
AgeMarriage	Age married; from age 11 to 23 or over, with 23 recording not married by age 22. Dependent variable in upper-limit censored Tobit.
LifeSatisfaction	Self-evaluated life satisfaction on a nine-point scale with 9 best possible life and 0 worst possible life.
Education (highest grade)	Educational attainment. Highest grade achieved at time surveyed; scale from 0 (none) to15 (post-graduate)
Subjective Health	Subjective health on a 3-point Likert scale with1 poor health, 2 average health, 3 good health (recorded age 8–12 by parent/main carer, thereafter by the child)
Age (months)	Age of individual in months.
Female	Female gender: 1 female, 0 male.
Location: Rural	Area of residence is rural: 1 rural; 0 urban.
Wealth index	Composite index of household wealth on a continuous scale between 0 and 1. Measures housing quality, access to services (electricity, piped drinking water and adequate fuels for cooking) and ownership of consumer durables (e.g. a television, automobile or mobile phone).
Household Head: Male	Gender of head of household: 1 male, 0 female.
Household Head Education: Basic	Educational attainment of head of household is basic (below secondary): 1 yes, 0 no (no education or at least secondary).
Household Head Education: Secondary and above	Educational attainment of head of household is at least secondary; 1 yes, 0 no.
Household size	Number of people living in the household.
Eldest older sibling: brother	Eldest older sibling is male; 1 yes, 0 no (eldest older sibling is female or no older siblings).
Eldest older sibling: sister	Eldest older sibling is female; 1 yes, 0 no
Family shock	Exposed to family shocks (illness or death of a family member) since last surveyed: 1 yes, 0 no)
Child’s religion	Childs religion; 1 Hindu, 0 all other religions
Child’s caste	Childs caste or ethnic group: SC scheduled castes; ST scheduled tribes including Adivasis; BC economically and socially backward classes; OC other/open category (general class)
Region	Agro-climatic region of residence: Telangana, Rayalaseema or Coastal Andhra Pradesh
Height for age	Standardised z-score (based on WHO standards)
BMI for age	Standardised z-score (based on WHO standards)

Table A2. Bias analysis for early marriage

Outcome	δ ($\beta =0,$ $R_{\max }^2=0.8$)	δ ($\beta =0,$ $R_{\max }^2$ = 1)
Education	105.732	23.875
	(−1.627, −1.473)	(−1.627, −1.432)
Subjective Health	−1.905^a	−1.508
	(−0.61, −0.151)	(−0.61, −0.192)

Using the method of Oster (2019) delta, δ, provides estimates of the magnitude of omitted variables bias such that the estimated average treatment effect of early marriage in OLS regressions for each outcome goes to zero ($\beta =0).$ Delta is estimated using psacalc in Stata (Oster, 2013). $R_{\max }^2$ specifies the maximum R² for the full model with all potential unobservables. Figures in parenthesis define the identified set ($\widehat{\beta }$ in the uncontrolled regression and the controlled hypothetical regression with δ = 1). The covariates in the controlled regressions are those used in the PSM (consistent with the estimates in Table 4).

^a The negative sign for δ means that to overturn the results the unobservables need to be correlated with early marriage in the opposite direction to the direction of the observable characteristics and early marriage (Evans, 2019). For illustration, δ = −1.508 ($R_{\max }^2$=1) implies that to eliminate the observed effect of early marriage on subjective health the effect of the unobservables needs to be 1.508 times the effect of the observed variables.

Oster, E. (2019). Unobservable selection and coefficient stability: Theory and evidence. Journal of Business & Economic Statistics, 37(2), 187–204. doi:10.1080/07350015.2016.1227711

 Web of Science ®Google Scholar

Oster, E. (2013). PSACALC: Stata module to calculate treatment effects and relative degree of selection under proportional selection of observables and unobservables. Statistical Software Components S457677, Boston College Department of Economics, revised 18 Dec 2016.

 Google Scholar

Evans, B. J. (2019). How college students use advanced placement credit. American Educational Research Journal, 56(3), 925–954. doi:10.3102/0002831218807428

 Web of Science ®Google Scholar

Data availability statement

The data that support the findings of this study are available in Young Lives https://beta.ukdataservice.ac.uk/datacatalogue/studies/study?id=7823&type=Data%20catalogue.