Positive Parenting, Adolescent Adjustment, and Quality of Adolescent Diet in Nine Countries

Abstract

Introduction

We sought to understand the relation between positive parenting and adolescent diet, whether adolescents’ internalizing and externalizing behaviors mediate relations between positive parenting and adolescent diet, and whether the same associations hold for both boys and girls and across cultural groups.

Methods

Adolescents (N=1,334) in 12 cultural groups in nine countries were followed longitudinally from age 12 to 15. We estimated two sets of multiple group structural equation models, one by gender and one by cultural group.

Results

Modeling by gender, our findings suggest a direct effect of positive parenting at age 12 on a higher quality diet at age 15 for males (β = .140 95% CI: .057, .229), but an indirect effect of positive parenting at age 12 on a higher quality diet at age 15 by decreasing externalizing behaviors at age 14 for females (β = .011 95% CI: .002, .029). Modeling by cultural group, we found no significant direct effect of positive parenting at age 12 on quality of adolescent diet at age 15. There was a significant negative effect of positive parenting at age 12 on internalizing (β = −.065 95% CI: −.119, −.009) and externalizing at age 14 (β = −.033 95% CI: −.086, −.018).

Conclusions

We founder gender differences in the relations among positive parenting, adolescents’ externalizing and internalizing behaviors, and adolescent diet. Our findings indicate that quality of parenting is important not only in promoting adolescent mental health, but potentially also in promoting the quality of adolescents’ diet.

Introduction

High and increasing prevalence of overweight and obesity in adolescents is a major global public health problem. Since the 1980s, the global prevalence of overweight and obesity in children and adolescents has increased by 47% (Ng et al., 2014). This trend is observed in both high-income countries and low- and middle-income countries, with few gender differences (Ng et al., 2014). Although the rate of increase of overweight and obesity has slowed in high-income countries, the rate of increase is accelerating for children and adolescents in low- and middle-income countries (Ng et al., 2014). Adolescents with obesity are more likely to have immediate and long-term health risks such as adult obesity, type 2 diabetes, and cardiovascular disease (Reilly & Kelly, 2011).

Adolescence is a time of physical, cognitive, and social development, and is a period when health-related behaviors may be susceptible to change. Additionally, changes in body composition, insulin sensitivity, and growth during pubertal maturation make adolescence a critical period for the development of overweight and obesity (Alberga, Sigal, Goldfield, Prud Homme, & Kenny, 2012). Likely due to hormonal changes that occur during puberty, girls are at an increased risk of acquiring excess weight compared to similar-aged boys during adolescence (Alberga et al., 2012). Data from the National Longitudinal Study of Adolescent Health revealed that for both Black and White adolescents, overweight adolescent girls were much more likely to become obese as young adults, compared to overweight adolescent boys. This gendered disparity in incidence among overweight adolescents was particularly high for Black adolescent girls (Robinson, Stevens, Kaufman, & Gordon-Larsen, 2010). Additionally, this large gender gap in obesity in Black young adults did not reduce when setting adolescent behaviors (family dinners, hours of television, playing sports) to be equal across boys and girls (Robinson et al., 2010).

Adolescents are vulnerable to a number of changes in their diet. Studies have found that because adolescents spend increasing time outside of the home, they typically consume more fast foods and sugar-sweetened beverages and fewer fruits and vegetables (Alberga et al., 2012). During 2015–2018, on average, 12–19 year-olds in the US got 16.7 percent of their calories on a given day from fast food (Fryar CD, Carroll MD, Ahluwalia N, 2020). This was significantly higher for girls (18.5 percent compared to 14.9 percent of for boys; Fryar CD, Carroll MD, Ahluwalia N, 2020). In the Global School-Based Student Health Survey, 10.3 percent of adolescents consumed fast food 4–7 days per week. This was lowest in the Americas (8.3 percent) and highest in Southeast Asia (17.7 percent). Additionally, fruit and vegetable consumption were negatively associated with fast food consumption, and soft-drink consumption was positively associated with fast food consumption (Li et al., 2020).

Changes in quality of diet in adolescence can be prevented through addressing the social environment, such as parenting (Fisher, Mitchell, Smiciklas-Wright, & Birch, 2002; Pearson, Atkin, Biddle, Gorely, & Edwardson, 2010; Pearson, Biddle, & Gorely, 2009; Vereecken, Legiest, De Bourdeaudhuij, & Maes, 2009). Food-related parenting practices, such as adopting controlling feeding practices to prevent children from over eating in order to prevent negative health consequences, may be related to both positive and negative dietary behaviors in adolescence. Additionally, children tend to model their parents’ food consumption, such as intake of fruits and vegetables (Patrick & Nicklas, 2005). However, much of the evidence on parenting in relation to children’s dietary behaviors comes from early childhood and not adolescence.

Beyond feeding practices and modeling eating behaviors, a few studies have found a relation between parenting style and quality of diet of adolescents. For example, children who described their parents as authoritative compared to indulgent, authoritarian, or neglectful, consumed more fruit and fewer unhealthy snacks per day and ate breakfast on more days a week (Baumrind, 1991; Kremers, Brug, De Vries, & Engels, 2003; Pearson et al., 2010). However, there is a gap in the literature on the family environment and the construct of positive parenting in relation to adolescent diet. Positive parenting is broader in scope than authoritative parenting and is the “continual relationship of a parent(s) and a child or children that includes caring, teaching, leading, communicating, and providing for the needs of a child consistently and unconditionally” (Seay, Freysteinson, & McFarlane, 2014).

There is also a gap in the literature on the potential gendered effects of parenting on adolescent diet. A 2009 systematic review on family correlates of fruit and vegetable consumption in children and adolescents found that only four of the 25 identified studies examined associations separately for girls and boys, which is an important limitation because information on gender may be central to tailoring interventions (Pearson et al., 2009). Additionally, all but one of study in the review used a cross-sectional design.

In a systematic review of studies of diet and mental health in children and adolescents, two of the three studies on the relation between internalizing and externalizing behaviors and child diet demonstrated significant associations between mental health and quality of child diet (O’Neil et al., 2014). For instance, in a longitudinal pregnancy cohort study in Western Australia, internalizing and externalizing scores were significantly positively associated with a Western dietary pattern in comparison to a healthier eating pattern in adolescents at age 14 (Oddy et al., 2009). Internalizing and externalizing behaviors were associated with eating takeaway foods, red meat, and confectionary foods. Additionally, improved behavioral scores were associated with a greater intake of leafy green vegetables and fruit (Oddy et al., 2009).

Positive parenting is related to lower levels of internalizing and externalizing among adolescents, which have been linked with poorer quality of adolescent diet. This potentially indicates that internalizing and externalizing problems could mediate the direct relation between positive parenting and quality of adolescent diet. A meta-analysis found that positive parenting was associated with less relational aggression in children and adolescents (Kawabata, Alink, Tseng, van Ijzendoorn, & Crick, 2011). However, a systematic review found inconsistent evidence for longitudinal associations between positive parenting in childhood and internalizing problems in adolescents, indicating the need for additional research (Clayborne et al., 2020).

There is a gap in knowledge about relations among parenting, adolescent mental health, and adolescent diet, particularly studies that disaggregate the findings by gender and studies in low-and middle-income countries. Current research about parenting and diet is mostly from cross-sectional studies conducted in high-income countries such as the United States and the Netherlands that do not examine the findings separately for boys and girls (Fisher et al., 2002; Kremers et al., 2003). Many cross-cultural parenting studies have found consistency in the relations between parenting and child outcomes rather than between-culture differences, including more within-culture variation than between-culture variation (Deater-Deckard et al., 2018; Rothenberg et al., 2020). However, we could identify no longitudinal cross-cultural studies on the relations among parenting, adolescent mental health, and adolescent diet. It is important to understand these predictors of adolescent diet in diverse cultures, particularly in low- and middle-income countries because child and adolescent eating patterns have been transforming in these countries, and compared to previous decades adolescents in these countries are now at a higher risk of overweight, obesity, and associated non-communicable diseases (Adair & Popkin, 2005; Ng et al., 2014; Ochola & Masibo, 2014).

Study Aims and Hypotheses

Adolescent diet.

A second aim was to determine if internalizing and externalizing behaviors mediate the association between positive parenting and quality of adolescent diet. The third aim was to examine whether these relations significantly differ by gender and by culture. Our conceptual model (Figure 1) predicts that, controlling for socioeconomic status (SES), child gender, prior adolescent internalizing and externalizing behaviors, and concurrent parent diet, age 12 positive parenting will positively predict quality of adolescent diet at age 15. We hypothesized that this effect would be mediated by adolescent adjustment (internalizing and externalizing behaviors) at age 14. We did not hypothesize differences by gender or cultural group but examined those relations.

Figure 1.

Conceptual model

Note: MR=Mother Report, FR=Father Report, and CR=Child Report

Method

Research participants were part of the Parenting Across Cultures (PAC) study, a longitudinal study started in 2008 with the recruitment of children (N = 1,334; Mage = 8.28 years, SD = 0.64 years in wave 1) and their mothers and fathers in nine countries China, Colombia, Italy, Jordan, Kenya, the Philippines, Sweden, Thailand, and the United States). Selected in proportion approximating the distribution of the student population in each recruitment site, students from both public and private schools were recruited through letters sent home with them. In each site, families participated in annual interviews after their initial recruitment. Measures to address the present research questions were administered in waves 5 to 8, when children were ages 12 to 15, on average. At age 15, 72% of the original sample provided data. Continuing participants did not differ from those who did not provide age 15 data on parent age, parent marital status, and number of children in the household, but did differ on child gender and parental education. The study was approved by Institutional Review Boards at universities in each country.

Procedure and Measures

Measures were translated and back translated and subjected to a process of cultural adaptation to ensure linguistic and conceptual equivalence of the measures. After parents provided informed consent and children provided assent, interviews were conducted face-to-face, over the telephone, or online. Participants were given modest compensation for their time.

Table 1 provides means, standard deviations, and sample sizes for each variable in each site. Table 2 provides the bivariate correlations with Bonferroni adjusted significance level across the variables used in the model for all countries combined.

Table 1.

Descriptive Statistics by Culture (mean, standard deviation, sample size)

Country/Cultural Group	Gender (Male)	SES (Age 15)	Child Diet Age 15	Parent Diet Age 15	Positive Parenting Age 12	Internalizing Age 14	Externalizing Age 14
China	49.59%	.74	.41	.59	−.48	−.51	−.62
		.72	.82	.87	1.02	.72	.67
	N=123	N=43	N=37	N=45	N-85	N=45	N=45
Colombia	44.44%	−.31	−.45	−.39	.16	.58	.50
		0.29	0.99	.86	1.07	1.05	1.08
	N=108	N=80	N=80	N=80	N=93	N=79	N=79
Italy
Naples	48.00%	−.42	−.06	.01	−.26	.26	.20
		0.60	1.12	1.04	.94	1.01	.82
	N=100	N=84	N=87	N=84	N=93	N=85	N=85
Rome	52.29%	.04	.28	.31	−.62	.26	.18
		0.62	0.95	1.06	.97	.97	.90
	N=109	N=100	N=100	N=100	N=98	N=101	N=101
Jordan	52.63%	−.46	−.28	−.37	−.03	.08	.59
		0.29	0.87	.74	1.16	1.07	1.31
	N=114	N=101	N=101	N=101	N=104	N=104	N=104
Kenya	40.00%	−.65	.14	.25	.29	.13	−.37
		0.44	0.90	.70	.82	.73	.76
	N=100	N=86	N=75	N=86	N=93	N=80	N=80
Philippines	50.83%	0.06	−.27	−.26	.23	.22	.24
		0.61	0.98	.99	.76	.92	.89
	N=120	N=87	N=87	N=87	N=.91	N=90	N=90
Sweden	51.16%	0.93	.42	.41	.19	−..55	−.41
		2.41	0.86	.71	.80	.81	.65
	N=129	N=79	N=72	N=78	N=83	N=92	N=92
Thailand	50.83%	−0.14	.10	.16	−.12	−.06	−.05
		0.58	0.83	.75	1.05	.77	.90
	N=120	N=83	N=83	N=83	N=100	N=85	N=85
United States
African American	48.04%	0.10	−.15	−.43	.30	−.47	−.25
		0.74	1.08	1.31	.88	.91	1.06
	N=102	N=89	N=86	N=89	N=90	N=89	N=89
European American	58.18%	.76	.30	.28	.28	−.05	−.22
		.55	1.03	1.05	.74	1.21	.95
	N=110	N=95	N=88	N=95	N=96	N=92	N=92
Latinx	47.47%	−.33	−.20	−.27	.11	−.11	−.23
		0.62	1.07	1.02	1.16	.95	.89
	N=99	N=80	N=63	N=65	N=76	N=67	N=67

Table 2.

Correlations (* if Bonferroni−corrected p-value <.05, and N=sample size)

	Female	SES	Child Diet Age 15	Parent Diet	Positive Parenting Age 12	Internalizing Age 12	Externalizing Age 12	Internalizing Age 14	Externalizing Age 14
Female	-
SES	−.03 N=993	-
Child Diet Age 15	.111* N=959	.159* N=947	-
Parent Diet Age 15	.026 N=993	.153* N=990	.342* N=948	-
Positive Parenting Age 12	−.019 N=1094	.033 N=941	.100 N=910	.063 N=941	-
Internalizing Age 12	.128* N=1094	−.172* N=941	−.154* N=910	−.145* N=941	−.246* N=1094	-
Eternalizing Age 12	−.054 N=1094	−.184* N=941	−.229* N=910	−.186* N=941	−.290* N=1094	.621* N=1094	-
Internalizing Age 14	.198* N=1009	−.148* N=939	−.138* N=917	−.125* N=939	−.230* N=958	.650* N=958	.364* N=958	-
Externalizing Age 14	.002 N=1009	−.150* N=939	−.223* N=917	−182* N=939	−.269* N=958	.463* N=958	.731* N=958	.582* N=1009	-

Positive parenting.

Positive parenting was measured by mother, father, and youth report when the youth participants were 12 years of age. The measure consisted of four items from the Oregon Youth Study rating how much time parents spend with the child and how much they engage in positive parenting behaviors such as spending time with their child doing something special that he/she enjoys (Capaldi & Patterson, 1989). Children reported on their mother and father separately. Three of the four items were measured on a 5-point response scale (1=never; 2=less than once a month; 3=about once a month; 4=about once a week; 5=almost every day). The remaining item was “how many days a week do you sit and talk with your son/daughter?” or “How many days a week does your mother/father sit and talk with you?” which was measured from 1–7. The responses were standardized before creating a composite measure of positive parenting, an average of mother, father, and child report (alpha=.80). There was no significant difference in the mean of positive parenting for males and females (males=.00, females=−.02; t(63), p=.53).

Adolescent adjustment.

Internalizing and externalizing behaviors were measured at ages 12 and 14 using the Child Behavior Checklist and Youth Self-Report. (CBCL & YSR; Achenbach, 1991). The CBCL and YSR have been translated into at least 64 languages, and published studies have used these measures with at least 50 cultural groups. The measures have been used previously in all of the countries involved in the present study. Mothers, fathers, and children rated each item (e.g., fearful or anxious; cruelty, bullying, or meanness to others) on a 3-point scale (1=not true; 2=somewhat or sometimes true; 3=very true or often true). Standardized summed scores for mothers, fathers, and youth were averaged to create composites for internalizing (alpha at age 12=.92; alpha at age 14=.93) and externalizing (alpha at age 12=.93; alpha at age 14=.94). Adolescent internalizing at age 14 was significantly higher for females (males=−.16, females=.15, t(−6.4), p<.001). However, adolescent externalizing was not significantly different by sex (males=−.01, females=−.01, t(−.07), p=.94).

Quality of adolescent diet.

Quality of diet was measured using four items from a Health Behavior scale that was collected from youth participants at age 15. The items were adapted from the European Health and Behavior Survey (Wardle & Steptoe, 1991). The items asked youth to identify how often they consume fruits, vegetables, soda, and fast food (0=none; 1=one to three times a week; 2=four to six times a week; 3=one to two times each day; 4=three to four times each day). The items on soda and fast food consumption were reverse coded, and we created a composite average score of child healthy diet (alpha=0.40). Quality of adolescent diet was significantly higher for females (males=−.12, females=.11, t(−3.47), p<.001).

Control variables.

We controlled for maternal and paternal diet, SES, and youth gender (except in gender multiple group model). When the youth participants were age 15, mothers and fathers were asked the same questions from the Health Behavior scale on the quality of their diet. We created an average score of mother diet quality (alpha=0.46) and father diet quality (alpha=0.42). A composite score of SES when the youth participants were 15 years old was created through an average of the standardized scores of mother’s education and father’s education, and the gross annual household income. The household income ranged from one to ten, represented in the local currency for each site.

Analysis Plan

All continuous variables were standardized to a grand mean of 0 and a SD of 1 to yield easily interpretable relations between predictors and outcomes. We first estimated a structural equation multiple group model by gender using Mplus version 8 (Muthén & Muthén, 2017). Full information maximum likelihood was used to account for data missing at random. Because we started with the theoretical perspective that the hypothesized relations should be universal and not different by gender, we estimated a model that held all of the paths to be equal across girls and boys. Good model fit is defined by a non-significant chi square test, CFI and TLI greater than or equal to 95, and RMSEA less than or equal to 0.06 (Hu & Bentler, 1999). If good model fit is not achieved, modification indices are then consulted to determine parameters that would be theoretically plausible to free across boys and girls. The parameter with the largest index is freed and no longer constrained to be equal across boys and girls. This model with the unconstrained parameter is then compared to the fully constrained model using a chi-square test. If the test reveals a significant difference in fit, then the unconstrained parameter is retained, modification indices are again consulted, and the plausible parameter with the largest index is freed. This iterative process was continued until the chi-square test comparing the constrained and the unconstrained model was not significant, indicating that the model freeing the parameter (less parsimonious model) fits the data significantly worse than the more parsimonious fixed model. The indirect effects of positive parenting on quality of adolescent diet and the corresponding bootstrapped 95% confidence intervals were calculated for the final model. Additionally, we conducted a multiple group model to test whether the model is different by PAC site, controlling for gender.

Results

Table 3 provides the unstandardized results for the final multiple group model for gender (controlling for PAC site using fixed effects), including bootstrapped confidence intervals and the indirect effect of positive parenting on quality of adolescent diet through child internalizing and externalizing. The model constraining all paths to be equal across girls and boys fit the data well (χ²(56)=78.807, p=.024, RMSEA=0.025, 90% CI=(0.009, 0.037), CFI=0.988, TLI=0.978). After freeing two paths to be different across boys and girls (2.27 percent of paths), the model fit was not significantly worse than the constrained model, (χ²(54)= 65.326, p = 0.139, RMSEA = 0.018, 90% CI = (0.000, 0.032), CFI = 0.941, TLI = 0.989). In this model, we found that positive parenting significantly predicted quality of adolescent diet in boys, but not girls (boys= 0.140, 95% CI: 0.057, 0.229; girls= 0.031 95% CI: −0.058, 0.123). In both boys and girls, externalizing at age 14 significantly negatively predicted quality of adolescent diet (−0.114, 95% CI= −0.200, −0.031), but internalizing at age 14 did not (−0.028, 95% CI=−0.099, 0.48). Additionally, although positive parenting at age 12 was significantly negatively related to internalizing at age 14 for both boys and girls (−0.064, 95% CI=−0.127, −0.005), it was only significantly related to externalizing among girls (boys: 0.023, 95% CI=−0.052, 0.101 girls: −0.092, 95% CI=−0.166, −0.023). We tested the significance of the indirect effect of positive parenting on quality of adolescent diet through child externalizing and internalizing behavior for boys and girls. We found a significant indirect effect of positive parenting on child diet through externalizing for girls (0.011, 95% CI=0.002, 0.029). However, there was no significant indirect effect of positive parenting through externalizing for boys (−0.003, 95% CI=−0.014, 0.005) and no significant indirect effect of positive parenting through internalizing for either boys or girls.

Table 3.

Unstandardized Results for Gender Multiple Group Model

	Std. Est	95% CI
Predictors of Age 15 Child Diet
Parent diet age 15	0.246	0.177	0.316
Positive parenting age 12	Male: 0.140 Female: 0.031	0.057 −0.058	0.229 0.123
Externalizing age 14	−0.114	−0.200	−0.031
Internalizing age 14	−0.028	−0.099	0.048
Socioeconomic status	0.086	0.012	0.216
Shanghai, China	−0.138	−0.438	0.193
Naples, Italy	−0.243	−0.526	0.021
Kenya	−0.232	−0.541	0.040
The Philippines	−0.468	−0.733	−0.207
Thailand	−0.217	−0.463	0.024
Sweden	−0.122	−0.388	0.142
US African American	−0.394	−0.688	−0.119
US European American	−0.150	−0.418	0.149
US Latinx	−0.428	−0.747	−0.116
Colombia	−0.548	−0.842	−0.261
Jordan	−0.340	−0.563	−0.112
Predictors of externalizing age 14
Externalizing age 12	0.702	0.640	0.763
Positive parenting age 12	0.023 Female: −0.094	−0.052 −0.166	0.101 −0.023
Socioeconomic status	−0.037	−0.091	0.024
Shanghai, China	−0.160	−0.385	0.048
Naples, Italy	−0.028	−0.200	0.141
Kenya	−0.193	−0.380	0.013
The Philippines	0.021	−0.154	0.185
Thailand	0.050	−0.127	0.216
Sweden	−0.051	−0.217	0.107
US African American	−0.212	−0.416	−0.004
US European American	−0.045	−0.257	0.157
US Latinx	−0.088	−0.273	0.114
Colombia	0.332	0.131	0.550
Jordan	0.208	−0.008	0.429
Predictors Internalizing age 14
Internalizing age 12	0.602	0.550	0.654
Positive parenting age 12	−0.064	−0.127	−0.005
Socioeconomic status	−0.051	−0.120	0.022
Shanghai, China	−0.226	−0.485	0.020
Naples, Italy	−0.130	−0.331	0.073
Kenya	−0.102	−0.321	0.111
The Philippines	−0.109	−0.315	0.098
Thailand	−0.177	−0.367	0.022
Sweden	−0.225	−0.436	−0.003
US African American	−0.368	−0.593	−0.148
US European American	−0.054	−0.258	0.170
US Latinx	−0.027	−0.238	0.196
Colombia	0.325	0.120	0.548
Jordan	−0.117	−0.350	0.124
Indirect Effect of positive parenting age 12 through:
Externalizing age 14	−0.003 Female: 0.011	−0.014 0.002	0.005 0.029
Internalizing age 14	0.002	−0.002	0.010

We also ran a multiple group model for PAC site (Table 4). Our initial multiple group model constraining all of the paths to be equal across sites fit the data well (χ²(224)=254.165, p=.081, RMSEA=.035, 90% CI=(0.000, 0.054), CFI=0.982, TLI=0.980). After freeing 2.38 (4) percent of all of the possible paths, the model fit significantly better (χ²(222)=233.602, p=.476, RMSEA=.005, 90% CI=(0.000, 0.040), CFI=1.000, TLI=1.000). Across all sites, positive parenting at age 12 was not significantly related to quality of adolescent diet (0.069, 95% CI=−0.004, 0.133). Across all sites, adolescent externalizing at age 14 was significantly negatively related to quality of adolescent diet. A one SD increase in child externalizing at age 14 was associated with a 0.108 SD decrease in the quality of adolescent diet (95% CI=−0.194, −0.024). Additionally, across all 12 sites, internalizing was significantly related to quality of adolescent diet. Positive parenting significantly negatively predicted internalizing at age 14 (−0.065, 95% CI=−0.119, −0.009) and externalizing at age 14 (−0.033, 95% CI=−0.086, −0.018). We tested the significance of the indirect effect of positive parenting on child diet through child externalizing and internalizing behavior. Across all sites, there were no statistically significant indirect effects (externalizing=0.004, 95% CI=−0.001, 0.013; internalizing=0.003, 95% CI=−0.001, 0.011).

Table 4.

Unstandardized Results for Parenting Across Cultures Site Multiple Group Model

	Std. Est	95% CI
Predictors of Age 15 Child Diet
Parent diet age 15	0.252 Kenya: −0.118	0.181 −0.351	0.319 0.107
Positive parenting age 12	0.069	−0.004	0.133
Externalizing age 14	−0.108	−0.194	−0.024
Internalizing age 14	−0.040	−0.115	0.029
Socioeconomic status	0.083	0.017	0.144
Female	0.220	0.100	0.351
Predictors of externalizing age 14
Externalizing age 12	0.729	0.676	0.781
Positive parenting age 12	−0.033	−0.086	−0.018
Socioeconomic status	−0.054	−0.098	−0.008
Female	0.156 Naples: −0.200 USAA: −0.253	0.065 −0.454 −0.488	0.260 0.056 −0.041
Predictors of internalizing age 14
Internalizing age 12	0.603	0.553	0.649
Positive parenting age 12	−0.065	−0.119	−0.009
Socioeconomic status	−0.039 Kenya: −0.257	−0.098 −0.408	0.019 −0.060
Female	0.247	0.155	0.350
Indirect Effect of positive parenting age 12 through:
Externalizing age 14	0.004	−0.001	0.013
Internalizing age 14	0.003	−0.001	0.011

Note: when a country is specified, it means that the model fit significantly better to allow the coefficient in that country to be different from the other countries.

Discussion

The aims of this research were to understand relations between positive parenting and adolescent diet and to test whether adolescents’ internalizing and externalizing behaviors mediate relations between positive parenting and adolescent diet and whether these associations hold across gender and across cultural groups, controlling for parents’ diet and other covariates. We addressed these aims in a sample of adolescents followed longitudinally from 12 to 15 years in 12 cultural groups in nine countries.

In the multi-group model testing for gender differences, we found a significant direct effect of positive parenting at age 12 on age 15-child diet for males. There was a significant negative effect of positive parenting at age 12 on internalizing at age 14 for boys and girls but a significant negative effect of positive parenting at age 12 on externalizing at age 14 only for girls. Only one indirect effect was significant: positive parenting at age 12 on quality of adolescent diet at age 15 through externalizing at age 14 for girls. The effect of positive parenting on child externalizing for boys and girls could be different during adolescence.

Examining the sample by culture, we found no significant direct effect of positive parenting at age 12 on quality of adolescent diet at age 15, but there were significant negative effects of positive parenting at age 12 on internalizing and externalizing at age 14. The majority of effects were not significantly different by culture.

Taken together, our findings suggest a direct effect of positive parenting at age 12 on a higher quality diet at age 15 for boys but an indirect effect of positive parenting at age 12 on a higher quality diet at age 15 by decreasing externalizing behaviors at age 14 for girls. Although there are only a few studies on child mental health predicting quality of child diet, in 4–12 year-old Australian children, conduct problems predicted lower food and vegetable consumption for girls but not for boys (Renzaho, Kumanyika, & Tucker, 2011). Because mean quality of adolescent diet was significantly better for girls, it is possible that positive parenting leads to a better quality diet for boys but has a weaker association among girls, who have a higher quality diet already. The exception to this finding is among girls who exhibit high levels of externalizing behaviors, who have been found to be more likely to have poorer diets compared to girls with lower levels of externalizing behaviors.

Strengths of this study are having tested for longitudinal associations between positive parenting and quality of adolescent diet in a set of diverse countries around the world. We were able to examine potential heterogeneity by gender and by culture. However, we must acknowledge some limitations. First, although the samples were designed to be representative of the cities from which they were drawn, they are not nationally representative, so findings may not generalize to entire countries included in this study. Within-country differences related to SES or region (e.g., urban vs. rural) are important considerations in generalizability. Second, because quality of adolescents’ and parents’ diet was only measured at age 15, we were not able to control for previous quality of children’s and parents’ diet when examining relations between positive parenting and quality of adolescent diet. Third, due to small sample sizes within countries, we were unable to estimate a model simultaneously stratifying the results by both cultural groups and gender. Therefore, we are unable to test whether the results in the gender multiple group model are consistent across cultures.

Conclusions

Taken together, the findings suggest that for girls, externalizing behavior mediates the relation between positive parenting and quality of adolescent diet, whereas positive parenting is directly related to increased quality of adolescent diet for boys, controlling for parents’ diet, previous behavior problems, and SES. We found little evidence for cultural differences in associations among positive parenting, adolescents’ externalizing and internalizing behaviors, and adolescent diet. Work is needed to understand the simultaneous heterogeneous relations among parenting, adolescent mental health, and quality of adolescent diet by both culture and gender together. However, our findings indicate that quality of parenting is important not only in promoting adolescent mental health, but potentially also in promoting quality of adolescent diet. These findings could be important in integrating parenting into interventions to promote healthy diet and prevent overweight and obesity in adolescents.

Data availability statement:

The data that support the findings of this study are available from the corresponding author upon reasonable request.