Page 34 - IJPS-8-1

International Journal of
Population Studies School dropouts and mental health among Indian adolescents

compare the mental condition of women with no child, age and was categorized in four groups “wife older than
with children < 2 years of age and children older than husband or wife 2 years younger than husband),” “wife
2 years of age. This variable is used to compare the impact three to 6 years younger to husband,” “wife more than
of raising younger children and children older than 2 years 6 years younger than husband,” and “do not know” (DK)
on the women’s mental health since younger children (husband’s age was not reported). The categorization of
require more attention and care from their mothers. spousal age gap is done based on the distribution of data.
Further, being younger than husband (more than 6 years)
2.2.3. Control variables from baseline is also a proxy indicator of low level of empowerment

Background characteristics variables of the females taken of young girls. Aforesaid variables – such as education,
from the baseline survey are as follows; Religion (Hindu and self-efficacy, decision-making power, working status, and
Non-Hindu), caste (scheduled caste/scheduled tribes, other spousal age gap are proxy indicators of empowerment of
backward class [OBC], and General), family type (Nuclear young women.
family and Non-nuclear family), parents’ education
(educated and not educated) parents’ working status (No and 2.3. Statistical Analysis
Yes), and parents’ life (One of the parents or both died and We used bivariate analyses (cross-tabulations and Chi-
both are alive). Data provided a household’s wealth index square tests) and ordered logistic regression (OLR) models
variable and constructed from the household’s conditions to examine school dropouts, early marriages, and early
and amenities (IIPS and Population Council, 2010). childbearing on mental health status at follow-up. OLR
is being used when the dependent variable is ordinal (i.e.,
2.2.4. Control variables from follow-up the variable has a meaningful order with more than two
Variables at the follow-up survey are also the key categories or levels). Here, the dependent variable (Mental
characteristics of the respondents that could be closely Health status) has three ordinal categories in nature, that
linked with their mental health. The following variables is, normal, moderate, and poor mental health status. The
are included respondent’s age (23–24, 25–26, and 26–27), poor mental health status is coded as the highest rank,
place of residence (rural and urban), household wealth whereas the normal is coded as the lowest rank. Three
index (Poor, Middle Rich), working status (Yes vs. No), self- multivariate models were fitted to understand the effects
efficacy (High vs. low), and decision-making power (Yes of sociodemographic variables measured in adolescence
vs. No). The self-efficacy is a variable that is determined by (Molde-1), marriage and childbearing in adolescence
combining two variables, expressing an opinion to older (Model-2), and variables measured in young adulthood
adults in the family, confronting if something went wrong, (Model-3) on mental health outcomes in young adulthood
that is, (1) whether the respondent is able to express an at follow-up. All analysis was carried out using STATA 15.0.
opinion (1-never, 2-sometimes, and 3-often) to older
persons in the family and (2) whether she confronts (1-stay 3. Results
quiet, 2-sometimes confront, and 3-always confront) The prevalence of women’s mental health status by GHQ-
if someone says or does something wrong to her. We 12 at baseline and follow-up surveys were shown in Table 1.
constructed self-efficacy score by summing the responses Cronbach alpha values suggested a higher consistency
to these questions and then dichotomized it as high versus in reporting depressive symptoms in both rounds. The
low. This question was asked by both unmarried and pattern of reported symptoms changed considerably over
married women. We considered four items relevant to the period. During adolescence, the statement “Not felt
decisions about matters related to their own lives, all are capable of making decisions” was reported by almost 21%
measured at follow-up – ‘decision in spending money, of adolescent girls, although such a proportion reduced to
about making major household purchases, about health merely 13% at follow-up. Similarly, the percentage of girls
care for herself and whether she should work or stay at who felt cannot overcome difficulties reduced from 15%
home. The score was assigned as follows: (1) Others only, to 12% over the cohort. On the other hand, to all other
(2) jointly with others, (3) alone, and an additive score for statements reporting of depressive symptoms increased
decision-making was constructed by summing the three over the period. For instance, girls who felt under strain
responses from the above-mentioned items. The higher increased from 6% in adolescence to 19% in follow-up,
the score, the higher the decision-making in adolescence. and those who were unhappy and depressed increased

Husband’s education (no education, primary, secondary, from 5% to 18%. Consistently, almost 10% of girls reported
higher secondary and above, DK) and spousal age gap were that they were not able to face up problems at both rounds
also modeled. Spousal age gap (at the follow-up survey) is of the survey. Overall, mental health status of adolescents
calculated by subtracting the wife’s age from the husband’s worsened from adolescence (15–19) to young adulthood

Volume 8 Issue 1 (2022) 28 https://doi.org/10.36922/ijps.v8i1.1280