Divorce and Eating Behaviors: A 5-Day Within-Subject Study of Preadolescent Obesity Risk

Abstract

Background:

Obesity rates have more than doubled among children and have tripled among adolescents since the 1980s, and currently more than one third of children and adolescents are overweight or obese. Parental divorce is a time of family upheaval, yet little is known about the family processes that link family structure and obesity.

Methods:

The current study gathered a 5-day eating behavior questionnaire from 37 preadolescents (mean=10.26 years; standard deviation=1.32; 32.4% female) and one parent to explore whether marital status was linked to obesity risk behaviors (i.e., high consumption of sugar-sweetened beverages (SSBs), low consumption of produce, skipping breakfast, and eating dinners away from the home) and whether family context (e.g., parent time spent with child, parental acceptance, and family routines) mediated that link.

Results:

Results showed that preadolescents in divorced families consumed more SSBs than preadolescents in married families, and there was a trend for less-frequent breakfast consumption among preadolescents in the divorced families. Of the three family context variables, only family routines explained the link between family structure and obesity risk.

Conclusions:

This study highlights the importance of family processes during divorce to understand the etiology and prevalence of child and adolescent obesity.

Introduction

Dramatic increases in the prevalence of obesity among youth populations in the United States and worldwide have become a major public health concern.^1,2 In the United States alone, 34.2% of 6- to 11-year-olds are overweight.³ Given that obese youth are at increased risk of becoming obese adults with serious physical and mental health issues, the identification of risk and protective factors that explain obesity are increasingly important.⁴ Because families are integral to eating behavior, childhood obesity is especially concerning for children who experience family transitions (particularly as they navigate through puberty).^5–7 The current study focuses on a specific transition event within some families (divorce) as a period of potential risk for behavioral patterns associated with later obesity.

Presently, nearly half of all marriages are predicted to end in divorce with over 1 million children each year experiencing a parental divorce.⁸ Divorce is problematic for families in a variety of ways, including: (1) disrupting parent-child relationships; (2) decreasing overall well-being for parents and children; (3) decreasing parental emotional support; (4) increasing economic adversity on the family; and (5) increasing disharmony between parents.⁹ In addition, and relevant to the obesity epidemic, unhealthy eating habits and poor physical health have been linked to divorce.^7,10,11 In contrast, healthy eating habits, such as regular consumption of produce, occur more often in two-parent families, suggesting that family context may protect against obesity risk.¹²

There are a number of known risk factors for childhood obesity. Currently, there is an abundance of data on children's anthropometric characteristics (e.g., physical measurements of body weight and height) and physical activity (PA). However, previous research has found that even after controlling for children's PA, divorce remained a significant predictor of children's high BMI.⁷ The present study focuses on four eating behaviors that have been linked to obesity risk: (1) consuming sugar-sweetened beverages (SSBs); (2) skipping breakfast; (3) eating less produce; and (4) dining outside the home.^13–16 Taken together, these four obesity risk behaviors likely create a compounding obesity risk index for children. The aim of this study was to examine the links between these obesity risk behaviors and family structure using reports of eating behaviors to gather across-day information from participants.

Parental influence and family functioning play a key role in children's development of healthy eating behaviors.^5,6 Specifically, parents model healthy eating behaviors and control children's food intake, diet quality, portion size, and food exposure in the home.⁵ Further, poor family functioning, including few routines and lack of communication between family members, has been linked to unhealthy eating behaviors among children. Families with children who are obese tend to report more family conflict, less family cohesion, and more difficulty managing family meals.^16,17 However, the direction of the relationship between childhood obesity and poor family functioning remains unclear. Owing to these findings on the effects of family context on eating behavior, it is important to focus on the family processes that might serve as a buffer against negative outcomes of divorce, such as obesity risk.^5,6,16,17

In addition to playing an important role in the overall health of youth, families are an important context for understanding children's and adolescents' well-being.¹⁸ Specifically, adolescents who perceive more parental rejection report more psychological disorders than adolescents who perceived less parental rejection.¹⁹ Further, adolescents from families with low connectedness report less parental acceptance and more parental rejection than adolescents from more-connected families. Because family context is linked to adolescent health and well-being and obesity prevention treatment programs have linked parental involvement to salutary outcomes, it was proposed in the current study, that protective aspects of the family may explain the association between divorce and childhood obesity.^18,20,21

Little is known about the relationship between specific eating habits and divorce, but given the instability resulting from divorce, it was anticipated that it would be a time of risk and that divorce would tend to be associated with obesity risk factors. Further, family context predictors of parental acceptance, time spent with parents each day, and family routines were linked between divorce and eating behavior outcomes. Thus, the current study explored whether qualities of the family context mediate the link between marital status and obesity risk behaviors in a sample of preadolescent youth. Based on previous literature, the following was hypothesized:

Hypothesis 1: Preadolescents from divorced families will report more obesity risk behaviors (i.e., more SSBs, fewer breakfasts, less produce, and fewer dinners home) than preadolescents from married families.

Hypothesis 2: The link between eating behaviors and divorce will be explained by the amount of time the preadolescent spends with the parent each day.

Hypothesis 3: Greater parental acceptance will mediate the relation between divorce and obesity risk.

Hypothesis 4: More family routines will mediate the relation between divorce and obesity risk.

Methods

Participants

Thirty-seven parent-preadolescent dyads were recruited from the San Francisco Bay Area. Twenty-three of the dyads were from two-parent coresident families and 14 were from families that were divorced or separated within recent years (mean [M] time since separation=2.56 years; standard deviation [SD]=2.27). Given the modest effect sizes typically associated with divorce and the repeated measurements, the sample size is adequate to detect divorce trends.²² Families were eligible if they were English speaking with one or more preadolescents between 8 and 12 years of age. If families had more than 1 eligible child, parents chose 1 child to participate. Parents included 32 females and 5 males (86.5% female) and ranged in age from 31.37 to 52.06 years, with an average age of 43.19 (SD=4.95). Preadolescents included 12 girls and 25 boys (32.4% girls) and ranged in age from 8.26 to 12.83 years, with an average age of 10.26 (SD=1.32). Parents provided open-ended responses regarding ethnicity, with 67.6% of participants reporting that they were of European ancestry, 16.2% of Asian ancestry, 10.8% of Latino ancestry, 2.7% of African ancestry, and 2.7% classified as multicultural/other ethnicity. See Table 1 for descriptive statistics of the sample.

Table 1.

Sample Descriptive Statistics

	Married families	Divorced families
	(n=23)	(n=14)
Parent age (SD)	43.06 (4.82)	43.38 (5.17)
Child age (SD)	9.90 (1.27)	10.87 (1.18)
Parent gender (%)
Male	2 (8.7)	3 (21.4)
Female	21 (91.3)	11 (78.6)
Child gender (%)
Male	16 (69.6)	9 (64.3)
Female	7 (30.4)	5 (35.7)
Parent's ethnicity (%)
Caucasian/white	16 (69.6)	9 (64.3)
Asian	4 (17.4)	2 (14.3)
Hispanic/Mexican American	2 (8.7)	2 (14.3)
African American	—	1 (7.1)
Multiracial/other	1 (4.3)	—
Child's ethnicity (%)
Caucasian/white	13 (56.5)	7 (50.0)
Asian	2 (8.7)	1 (7.1)
Hispanic/Mexican American	1 (4.3)	2 (14.3)
African American	—	—
Multiracial/other	7 (30.4)	4 (28.6)

SD, standard deviation.

Materials

A website was used by parents to assess their eligibility for this study. Following a telephone conversation, families received a packet in the mail that contained consent forms (for parent and preadolescent), a booklet with the 5-day eating behavior questionnaire (for parent and preadolescent), interview response cards (containing Likert response options for phone interview surveys), and return envelopes for the materials. The packets also included a chart detailing serving-size exchanges for different types of produce.

Procedure

Divorcing families were recruited from San Francisco Family Court records (n=8) and married families (and other divorced/separated families) were recruited through (1) Parent Teacher Association meetings (n=1) and (2) online postings (n=27) while the recruitment method for one family was unknown. Eligible parents participated in a 10-minute telephone prescreening interview during which they learned about the study, chose one preadolescent to participate, and scheduled a time for two separate telephone interviews (one for the parent and one for the preadolescent). Consent forms, response option index cards, and eating booklets were then mailed to families. Next, an interviewer called each family and completed separate telephone interviews with the parent and the preadolescent, each lasting approximately 45 minutes. At the end of the interviews, each parent-preadolescent dyad was trained to complete the booklet that included the 5-day eating behavior questionnaire. In each booklet, across all 5 days, preadolescents responded to a series of items regarding the food they ate throughout each day. At the end of the 5-day study, participants mailed the materials back to the research team. Participants received either a $30 money order or a $30 gift certificate as compensation for their participation in the telephone interview and received an additional $10 money order or $10 gift certificate for completion of the 5-day eating behavior questionnaire. All study procedures and instruments were approved by the San Francisco State University Institutional Review Board.

Measures

Five-day eating behavior questionnaire

To obtain a dynamic measure of eating habits, parents and preadolescents responded to questions related to daily eating habits across 5 days. Data were collected on nonconsecutive days, including at least one weekend day. Parents and preadolescents were asked about four obesity risk behaviors: (1) consumption of SSBs; (2) eating breakfast; (3) consumption of produce; and (4) dinners eaten at home. To assess the number of SSBs consumed, parents and preadolescents were asked to report for each day: “How many servings of sugar-sweetened drinks have you consumed today? (e.g., soda, Gatorade, fruit punch).” Participants were asked to respond using a Likert-type scale ranging from 0 (no servings) to 8 (eight or more servings). To assess whether breakfast was eaten, parents and preadolescents were asked to report for each day: “Did you eat breakfast today?” Response options were 1 (yes) or 2 (no). To assess produce consumption, parents and preadolescents were asked to report for each day: “How many servings of produce (fruits, vegetables) have you eaten today?” Participants were asked to respond using a Likert-type scale that ranged from 0 (no servings) to 8 (eight or more servings).

Family structure

Parents were asked questions concerning family structure, including marital status, number of children, and ages of each child, during the initial interview. For marital status, families that were married or cohabitating were assigned to the married group and families that were divorced or separated were assigned to the divorced group.

Family routines

To assess family routines, seven items from the 23-item Family Routines Inventory were used.²³ The seven items were chosen because they are indicators of family routines that are particularly affected by divorce, including spending time with the child and sharing activities.²⁴ Parents responded to the items on a 3-point Likert scale that ranged from 1 (never) to 3 (always). A composite measure of the seven items was estimated so that higher scores indicated more consistent family routines. The reliability for the composite of these seven items was α=0.74.

Parent acceptance

The parent acceptance subscale of the Children's Report of Parent Behavior Inventory assessed the preadolescents' perceptions of parental acceptance.²⁵ The 16-item subscale asked preadolescents whether specific behaviors were 1 (not like), 2 (somewhat like), or 3 (like) their own parent's behavior (e.g., “Your parent seemed to see your good points more than your faults.”). Higher scores indicated higher levels of perceived parental acceptance, and an average was computed across the scores. The reliability for this subscale was α=0.78

Parent time spent with child

Preadolescents were asked the following open-ended question on each of the 5 days: “How many hours were you awake and with your parent who is participating with you?” The 5-day eating booklets remained with the preadolescents regardless of whether they were spending time with the participating parent or nonparticipating parent. In addition, parents were asked the following open-ended question: “Out of a total of 100%, what percentage did your child spend with you or in your care?” Higher scores indicated more waking hours spent together.

Results

Descriptive Statistics

Preadolescents reported relatively healthy eating behaviors overall. Across the 5 days, preadolescents from both married and divorced families reported eating breakfast greater than 94% of the time and eating three or more servings of produce per day. The number of SSBs consumed was relatively low for preadolescents in both the married (M=0.53) and divorced groups (M=1.18). Parents from the married families reported a greater frequency of family routines, when compared to parents from the divorced families (t[32]=2.15; p=0.04). See Table 2 for descriptive statistics of outcome variables.

Table 2.

Descriptive Statistics for Obesity Risk Behaviors and Family Context Indicators by Family Type

	Married families				Divorced families
	Children		Parents		Children		Parents
	M	SD	M	SD	M	SD	M	SD
Servings of produce per day	3.65	1.70	3.57	1.65	4.03	1.66	3.99	1.49
Servings of SSBs per day	0.53	1.11	0.13	0.36	1.18	0.14	0.70	1.15
% of breakfasts eaten	94.55		95.65		96.88		85.71
% of dinners eaten at home	85.32		78.95		84.62		85.71
Hours parent and preadolescent spent together per day^a	9.27	3.27			7.56	4.60
Parental acceptance^a	2.75	0.19			2.71	0.32
Frequency of family routines^b			2.69	0.17			2.53	0.26

The only significant difference between married and divorced families was in frequency of family routines (p=0.040).

Reported by the child.

Reported by the parent.

SSBs, sugar-sweetened beverages; M, mean; SD, standard deviation.

Tests of Hypotheses

Because parents and preadolescents were interviewed to gather information about family context and measured obesity risk behaviors from multiple-day questionnaires of eating behaviors, the current study includes a great deal of correlated data. Therefore, a mixed-model design was used for analysis.²⁶ A mixed-model design combines the between-group factor of family structure with the within-subject factors of family context simultaneously between groups and over time for the obesity risk behaviors.²⁷

Initial model of change over 5 days

First, we assessed whether there was a significant change over the 5 days of the study for any of the obesity risk behaviors (e.g., consumption of SSBs, breakfast, produce, or dinners home). However, all p values were nonsignificant (all p values >0.323), suggesting reports on the obesity risk factor behaviors did not change as a function of participation in the study. See Tables 3 and 4 for statistical estimates for SSBs and breakfast consumption.

Table 3.

Parameter Estimates of the Fixed and Random Effects of the Conditional and Intercept-Only Models Concerning the Effects of Marital Group, Parent Time Spent with Child, Parental Acceptance, and Family Routines on Child Obesity Risk Factors (SSBs and Breakfast)

	SSBs				Breakfast
			95% CI				95% CI
Parameters	Estimate		Lower	Upper	Estimate		Lower	Upper
Initial model
Intercept	1.14	^***	0.67	1.61	1.03	^***	0.94	1.13
Marital group (0=married; 1=divorced)	−0.66	^*	−1.20	−0.13	−0.07		—	—
Child sex (1=female; 2=male)	0.01		—	—	−0.04		—	—
Parent time with child
Intercept	1.17	^***	0.69	1.64	1.03	^***	0.93	1.13
Marital group (0=married; 1=divorced)	−0.68	^*	−1.22	−0.14	−0.07		—	—
Child sex (1=female; 2=male)	0.04		—	—	−0.04		—	—
Parent time with child	0.00		—	—	0.00		—	—
Parental acceptance
Intercept	4.28	^**	1.17	7.39	0.88	^**	0.23	1.54
Marital group (0=married; 1=divorced)	−0.63	^*	−1.16	−0.10	−0.08		—	—
Child sex (1=female; 2=male)	−0.21		—	—	−0.03		—	—
Parental acceptance	−1.13	^*	−2.24	−0.03	0.05		—	—
Family routines
Intercept	4.27	^**	1.07	7.46	0.57	^*	0.07	1.06
Marital group (0=married; 1=divorced)	−0.23		—	—	−0.05		—	—
Child sex (1=female; 2=male)	−0.27		—	—	−0.01		—	—
Family routines	−1.28	^*	−2.55	−0.02	0.16		—	—

p<0.05; ^**p<0.01; ^***p<0.001.

SSBs, sugar-sweetened beverages; CI, confidence interval.

Table 4.

	Produce				Dinners home
			95% CI				95% CI
Parameters	Estimate		Lower	Upper	Estimate		Lower	Upper
Initial model
Intercept	4.24	^***	3.56	4.92	1.15	^***	1.05	1.26
Marital group (0=married; 1=divorced)	−0.38		—	—	−0.02		—	—
Child sex (1=female; 2=male)	−0.49		—	—	0.04		—	—
Parent time with child
Intercept	4.24	^***	3.55	4.93	1.15	^***	1.05	1.26
Marital group (0=married; 1=divorced)	−0.38		—	—	−0.02		—	—
Child sex (1=female; 2=male)	−0.49		—	—	0.04		—	—
Parent time with child	0.00		—	—	0.00		—	—
Parental acceptance
Intercept	3.07		—	—	0.35		—	—
Marital group (0=married; 1=divorced)	−0.37		—	—	−0.04		—	—
Child sex (1=female; 2=male)	−0.43		—	—	0.12		—	—
Parental acceptance	0.42		—	—	0.29	^*	0.05	0.52
Family routines
Intercept	−2.31		—	—	1.02	^*	0.24	1.80
Marital group (0=married; 1=divorced)	−0.94	^*	−1.83	−0.05	−0.04		—	—
Child sex (1=female; 2=male)	−0.49		—	—	0.03		—	—
Family routines	2.64	^**	0.71	4.56	0.06		—	—

p<0.05; ^**p<0.01; ^***p<0.001.

CI, confidence interval.

Marital status

Next, we estimated whether the fixed factor of family structure explained differences in reports of each obesity risk behavior. Results showed that preadolescents in divorced families consumed more SSBs than preadolescents in married families (t[53.24]=−2.49; p=0.016; 95% confidence interval [CI], −1.20, −0.13), and there was a trend linking eating breakfast to marital status (p=0.168). There was no significant difference between marital groups in amount of produce consumed (p=0.324) or number of dinners eaten at home (p=0.743).

Parent time spent with child

Next, we tested whether adding the family context predictors diminished the link between obesity risk behaviors and family structure. First, we tested whether the amount of time parents spent with their child served as a mediator, but time together did not reduce the link between divorce and consumption of SSBs, eating breakfast, produce consumption, or dinners eaten at home (all p values >0.477).

Parental acceptance

Next, parental acceptance was tested as a mediator and found to be significantly related to consumption of SSBs (t[51.80]=−0.2.05; p=0.045; 95% CI, −2.24, −0.03) and to dinners home (t[61.14]=2.43; p=0.018; 95% CI, 0.05, 0.52), but that it did not mediate the link between divorce and either risk factor. Parental acceptance was not related to eating breakfast (p=0.639) or produce consumption (p=0.601).

Family routines

Finally, frequency of family routines was tested as a mediator and found to be significantly related to consumption of SSBs (t[49.96]=−2.03; p=0.047; 95% CI, −2.55, −0.02) and mediated the link between divorce and consumption of SSBs (t[50.07]=−0.80; p=0.429). Family routines was also significantly related to consumption of produce (t[41.82]=2.77; p=0.008; 95% CI, 0.72, 4.56), but did not mediate the relationship between divorce and produce consumption. Frequency of family routines was not related to eating breakfast (p=0.109) or dinners home (p=0.705).

Discussion

In summary, preadolescents in the divorced families appeared to consume more SSBs when compared to preadolescents in the married families. In addition, a trend of eating breakfast less frequently was found for preadolescents in the divorced families when compared to preadolescents in the married families. Contrary to expectations and previous research, no group differences were found for consuming produce and eating dinners at home.¹² In contrast with the hypothesis, time parents spent with their preadolescents did not mediate the relationship between divorce and any of the obesity risk factors. Interestingly, parental acceptance was associated with greater consumption of SSBs and dinners eaten home, but it did not influence the relationship between divorce and either risk factor. Last, greater frequency of family routines was found to be protective against consumption of SSBs and mediated the relationship between divorce and SSBs. Greater frequency of family routines was also related to greater produce consumption, but did not mediate the relationship between divorce and produce consumption.

The finding that more family routines were related to higher produce consumption suggests that more consistent family routines and rituals may provide a stable family environment that helps foster positive parent-child relationships and healthy eating behaviors. Surprisingly, there was no association between family routines and breakfast consumption or dinners eaten at home. The possibility also exists that breakfast consumption and eating dinners at home may be predictive of time shared with the preadolescent, but not necessarily predictive of other family routines. The lack of associations between family routines and breakfast consumption and dinners eaten at home suggest that, regardless of whether preadolescents are eating meals with their parents, family routines unrelated to food, such as consistent bedtime, still protect against obesity risk behaviors.

That parental acceptance was linked to higher consumption of SSBs was an interesting finding. One interpretation is that preadolescents who do not have food restrictions put in place by their parents may perceive more parental acceptance than preadolescents whose parents monitor and correct unhealthy eating habits. In addition, parents may be more accepting of their preadolescents' poor eating behaviors when they mirror their own poor eating behaviors. An alternative explanation for why time spent with preadolescents did not mediate the link between divorce and obesity risk is that divorce and eating habits may not be equally linked to all forms of family functioning. In addition, divorce may reduce time spent with the preadolescent, but these findings suggest that children of divorce may be protected from obesity risk factors when family routines remain stable. Future research should continue to explore the mediating effect of family routines.

Owing to the preliminary nature of this research, the current study has several limitations. First, because of the limited size of the sample, there was a lack of statistical power to observe all the significant mediating relationships that might exist between divorce and the obesity risk behaviors. However, even with a small sample size and limited power, through the multiple-day design, the current study demonstrated statistically significant trends for breakfast consumption and SSBs. Additionally, given that the sample was relatively healthy overall, the power to detect the mediation model may have been underestimated owing to the limited range of reported eating behaviors.

Second, for the purposes of this study, only four eating behaviors that have been linked to obesity were examined; however, future research may continue to look into other obesity risk factors, such as family history of high BMI, parents' BMI, preadolescents' anthropometry (e.g., physical measurements of body weight and height), and preadolescents' level of PA. Also, previous research has shown that even after controlling for PA and socioeconomic differences, divorce was a significant predictor of high BMI among a sample of preadolescents.⁸ Further, there is no guarantee that parents or preadolescents followed their training and completed their eating booklets over the course of 5 separate days. Future researchers might time stamp booklets to avoid falsification.

Third, the current study did not analyze what the preadolescents ate or whether they were reporting the foods that they ate accurately. Future research should use additional methods for collecting eating data, such as devices used for dietary food analysis or hiring nutritionists to analyze pictures of food taken by participants using a camera phone.

Last, individual characteristics of the child were not controlled in the current study. Future research should examine how personal characteristics of the child, such as emotion regulation, temperament, and anxiety, might influence obesity risk. Though divorce can lead to negative outcomes for some children, others who undergo divorce are able to cope well. Research focused on characteristics related to resilience and positive health outcomes for children after divorce would be informative in this area. Additionally, the current study did not take into consideration the role of race and ethnicity. Given that previous research has shown that preadolescent obesity risk may be higher in African American and Latino families, future research should consider the potential role of parents' and preadolescents' racial and ethnic backgrounds.²⁸

Conclusions

In conclusion, it appears that family routines matter for eating behaviors and may potentially be a protective factor against obesity risk behaviors (such as consumption of SSBs) even when there are high levels of stress and conflict in the family surrounding a marital separation.²⁹ Future research should continue to explore this possibility and also examine how other areas of family functioning might influence the relationship between divorce and obesity risk behaviors.

These findings may have important implications for family-based intervention and prevention strategies. Because the consumption of food is a basic human need and because family context and divorce seem to be linked to eating behaviors, divorcing families may benefit from guidance regarding how they can integrate more family routines into their daily lives. These findings are a first step toward understanding how family processes and interactions contribute to the development of healthy or unhealthy eating behaviors. Future research might expand upon these findings by using more ethnically diverse samples and longitudinal research methods assessing family dynamics and eating behaviors during different developmental periods. Overall, the current study highlights the importance of family context (e.g., family routines) in understanding the etiology and prevalence of child and adolescent obesity.

Footnotes

Acknowledgments

The first and second authors contributed equally to the preparation of this article. The authors acknowledge and thank the families that devoted their time for this study. The authors also thank the members of the Family Interaction Research Lab for assisting with recruitment and analyses of these data, which made this work possible.

Author Disclosure Statement

No competing financial interests exist.

References

Harper

. Childhood obesity: Strategies for prevention. Fam Community Health, 2006; 29:288–298.

Wang

, Lobstein

. Worldwide trends in childhood overweight and obesity. Int J Pediatr Obes, 2006; 1:11–25.

Ogden

, Carroll

, Kit

, et al. Prevalence of childhood and adult obesity in the United States, 2011–2012. JAMA, 2014; 311:806–814.

Tiffin

, Arnott

, Moore

, et al. Modelling the relationship between obesity and mental health in children and adolescents: Findings from the Health Survey for England 2007. Child Adolesc Psychiatry Ment Health, 2011; 5:31.

Rhee

. Childhood overweight and the relationship between parent behaviors, parenting style, and family functioning. Ann Am Acad Pol Soc Sci, 2008; 615:11–37.

Savage

, Fisher

, Birch

. Parental influence on eating behavior: Conception to adolescence. J Law Med Ethics, 2007; 35:22–34.

Yannakoulia

, Panagiotakos

, Pitsavos

, et al. Eating patterns may mediate the association between marital status, body mass index, and blood cholesterol levels in apparently healthy men and women from the ATTICA study. Soc Sci Med, 2008; 66:2230–2239.

Amato

. The consequences of divorce for adults and children. J Marriage Fam, 2000; 62:1269–1287.

Wolchik

, Schenck

, Sandler

. Promoting resilience in youth from divorced families: Lessons learned from experimental trials of the New Beginnings Program. J Pers, 2009; 77:1833–1868.

10.

Aldous

, Ganey

. Family life and the pursuit of happiness: The influence of gender and race. J Fam Issues, 1999; 20:155–180.

11.

Fabricius

, Luecken

. Post-divorce living arrangements, parent conflict, and long-term physical health correlates for children of divorce. J Fam Psychol, 2007; 21:195–205.

12.

Friel

, Newell

, Kelleher

. Who eats four or more servings of fruit and vegetables per day? Multivariate classification tree analysis of data from the 1998 Survey of Lifestyle, Attitudes and Nutrition in the Republic of Ireland. Public Health Nutr, 2005; 8:159–169.

13.

Coppinger

, Jeanes

, Hardwick

, et al. Body mass, frequency of eating breakfast and consumption in 9–13-year-olds. J Hum Nutr Diet, 2012; 25:43–49.

14.

Huang

, Hu

, Fan

, et al. Associations of breakfast skipping with obesity and health-related quality of life: Evidence from a national survey in Taiwan. Int J Obes, 2010; 34:720–725.

15.

Stephens

, Summar

. Dietary interventions for the treatment of obese children and adolescents. Prog Pediatr Cardiol, 2008; 22:137–141.

16.

Beck

, Terry

. A comparison of obese and normal-weight families' psychological characteristics. Am J Fam Ther, 1985; 13:55–59.

17.

Jacobs

, Fiese

. Family mealtime interactions and overweight children with asthma: Potential for compounded risks?. J Pediatr Psychol, 2007; 32:64–68.

18.

Hammons

, Fiese

. Is frequency of shared family meals related to the nutritional health of children and adolescents?. Pediatrics, 2011; 127:e1565–e1574.

19.

Dwairy

. Parental acceptance-rejection: A fourth cross-cultural research on parenting and psychological adjustment of children. J Child Fam Stud, 2010; 19:30–35.

20.

Kitzman-Ulrich

, Wilson

, St. George

, et al. The integration of a family systems approach for understanding youth obesity, physical activity, and dietary programs. Clin Child Fam Psychol Rev, 2010; 13:231–253.

21.

Van Ryzin

, Nowicka

. Direct and indirect effects of a family-based intervention in early adolescence on parent-youth relationship quality, late adolescent health, and early adult obesity. J Fam Psychol, 2013; 27:106–116.

22.

Liu

, Liang

. Sample size calculations for studies with correlated observations. Biometrics, 1997; 53:937–947.

23.

Jensen

, James

, Boyce

, et al. The Family Routines Inventory: Development and validation. Soc Sci Med, 1983; 17:201–211.

24.

Vélez

, Wolchik

, Tein

, et al. Protecting children from the consequences of divorce: A longitudinal study of the effects of parenting on children's coping processes. Child Dev, 2011; 82:244–257.

25.

Schaefer

. A configurational analysis of children's reports of parent behavior. J Consult Psychol, 1965; 29:552–557.

26.

Bryk

, Raudenbush

. Hierarchical Linear Models: Applications and Data Analysis Methods. Thousand Oaks, CA: Sage, 1992.

27.

Heck

, Thomas

, Tabata

. Multilevel and Longitudinal Modeling with IBM SPSS. New York: Routledge/Taylor & Francis Group, 2010.

28.

Ogden

, Carroll

, Kit

, et al. Prevalence of obesity and trends in body mass index among US children and adolescents, 1997–2010. JAMA, 2012; 307:483–490.

29.

Riggio

. Parental marital conflict and divorce, parent-child relationships, social support, and relationship anxiety in young adulthood. Pers Relatsh, 2004; 11:99–114.