Behavioral Self-Regulation and Weight-Related Behaviors in Inner-City Adolescents: A Model of Direct and Indirect Effects

Abstract

Background: this study examined the association of two distinct self-regulation constructs, effortful control and dysregulation, with weight-related behaviors in adolescents and tested whether these effects were mediated by self-efficacy variables.

Methods: A school-based survey was conducted with 1771 adolescents from 11 public schools in the bronx, New york. self-reg-ulation was assessed by multiple indicators and defined as two latent constructs. Dependent variables included fruit/vegetable intake, intake of snack/junk food, frequency of physical activity, and time spent in sedentary behaviors. structural equation modeling examined the relation of effortful control and dysregulation to lifestyle behaviors, with self-efficacy variables as possible mediators.

Results: study results showed that effortful control had a positive indirect effect on fruit and vegetable intake, mediated by self-efficacy, as well as a direct effect. effortful control also had a positive indirect effect on physical activity, mediated by self-efficacy. Dysregulation had direct effects on intake of junk food/snacks and time spent in sedentary behaviors.

Conclusions: these findings indicate that self-regulation characteristics are related to diet and physical activity and that some of these effects are mediated by self-efficacy. Different effects were noted for the two domains of self-regulation. Prevention research-ers should consider including self-regulation processes in programs to improve health behaviors in adolescents.

Introduction

the epidemic of overweight in children observed in recent years has underscored the need for a better understanding of dietary and physical activity patterns in youth. the factors associated with obesity are multiple.

increases in total energy intake and sedentary activities, as well as consumption of high -density/low-nutritional-value foods, are lifestyle factors that have been suggested as related to the rising prevalence of obesity in children and adolescents.1–5 increasing intake of fruit and vegetables is a strategy for achieving a healthy weight, but recent reports indicate that there has been a decline in fruit and vegetable intake among adolescents over the years,6 and adolescents still consume less than the recommended daily servings of fruit and vegetables.6–8 Low-income and minority youth are the most affected by the obesity epidemic.9,10 thus, study-ing factors relevant to dietary choice and the development of overweight in this population offers the potential to better understand health disparities in youth.

Self-Regulation Abilities, Obesity, and Lifestyle Behaviors

self-regulation involves sets of related characteristics that develop as the child matures from early childhood to mid and late adolescence. Developmental studies indicate that two distinct processes are involved in self-regulation. this dual process model posits that risk derives from a reactive process whereas protection or resilience is offered by a more reasoned, thoughtful path. the dual-process model is supported by neuroimaging studies that describe the reactive processes (e.g., impulsivity) as being generated in the amygdala and related posterior areas of the brain, whereas the reasoned processes are based on regions of the prefrontal cortex.11–13 studies also suggest that the reactive/impulsive processes develop very early during infancy, whereas the reasoned/effortful control processes develop later during the preschool and school years and into late adolescence. Dysregulation includes characteristics that produce a pattern of responding to sit-

• 1Department of Epidemiology & Population Health, Albert Einstein College of Medicine, Bronx, NY.
• 2Prevention & Control Program, Cancer Research Center of Hawaii, Honolulu, HI.

• Preliminary findings were reported as a poster at the 30th Annual Meeting & Scientific Sessions of the Society of Behavioral Medicine, April 22–25, 2009, Montreal, Quebec, Canada.

uations that is reactive and impulsive, with less attention to future consequences of behavior.14–18 effortful control, on the other hand includes the abilities to inhibit or acti-vate behaviors for the benefit of long-term goals.

Currently, there is mounting evidence relating aspects of self-regulation to obesity and weight-related behaviors. seeyave et. al.19 showed that children with poor delay of gratification at age 4 had an increased risk of becoming overweight by the time they were 11 years old. Francis, et al.,20 using data from a large national sample, found that children with lower self-regulation skills (low self-control and poor delay of gratification) had a more rapid weight gain from 3 to 11 years old than children with better regulatory capacities.20 studies also showed that self-regulation skills are associated with dietary intake. A cross-sectional study among 4th grade students showed that effortful control skills were inversely related to snack food intake but not to fruit/vegetable intake.21 More recently, Riggs, et al.22 used multiple indicators for assess-ing effortful control in a prospective study of 4th grade students. this study showed that baseline effortful control predicted higher intake of fruit and vegetables and greater physical activity 4 months later. inverse association of effortful control has been reported for sedentary behav-iors.23 Although these studies used different methods for the assessment of self-regulation/effortful control, their results are consistent, indicating that the abilities acquired during the developmental process of self-regulation are relevant for obesity risk and weight-related lifestyle behaviors. However, the majority of these studies focused only on effortful control characteristics; few studies addressed reactive processes (impulsivity) in the context of obesity risk.24–26

Social-Cognitive Theory and Self-Control

social-cognitive theory proposes a theoretical link between self-regulation and self-efficacy: Aspects of effortful control such as problem solving, self-monitoring, and self-regulatory strategies such as goal setting have been proposed as being related to self-efficacy, in that individuals with better regulatory skills will feel more efficacious and more likely to succeed in achieving their behavioral goals.27–29As discussed above, prior studies have found that effortful control is related to healthy lifestyle behaviors (fruit/vegetable intake and physical activity);21–23,30 however, these studies did not address the role of self-efficacy in these associations. the empiri-cal evidence linking self-regulation and self-efficacy in the context of weight-related behaviors comes from a couple of studies by Anderson, et al.,31,32 who examined the effects of self-efficacy and the use of self-regulation strategies (i.e ., have written physical activity plans, have plans for bad weather, park further away to walk) in rela-tion to physical activity among older adults. Using struc-

tural equation modeling, they found that self-efficacy had a direct effect on outcomes and also had indirect effects through self-regulation strategies and outcome expectan-cies.31 in a second paper, these investigators reported that self-efficacy was related to fruit and vegetable intake and inversely related to fat intake.32 similarly, Annesi, et al. showed that among adults participating in a weight loss program increases in the use of self-regulatory strate-gies were related to increases in self-efficacy for eating healthy and being more physically active. this in turn was associated with weight loss but not with improve-ments in lifestyle behaviors.33,34 thus, despite the fact that these studies did not directly measure effortful control, their findings provide some empirical evidence of the interrelations of self-efficacy– and self-regulation–related capacities.

Present Research

the theoretical goal of this study was to test whether effortful control is related to diet and physical activity behaviors in a sample of inner-city adolescents, a group that is at higher risk of obesity and its complications. A second goal of the study was to test whether these effects were mediated by self-efficacy. the research was based on the proposition that effortful control is based on thoughtful processing of information and systematic decision making, as well as the regulation of impulse and emotion,35 which will contribute to the development of self-efficacy.36,37 Having greater cognitive resources may increase adolescents' abilities to employ self-regu-latory strategies, such as planning and problem solving, which in turn could translate into greater confidence of accomplishing behavioral goals. effortful control skills, therefore, could be important for building what bandura termed self-regulatory efficacy, the beliefs people have that they can motivate and regulate their own behaviors, which is critical for sustaining behavioral goals.27 How self-regulation and self -efficacy work in the context of social cognitive theory is further discussed by Zimmer-man38 and schunk.39 effortful control abilities, such as attention focusing and self-praise, are needed for chil-dren to feel more confident about accomplishing their behavioral goals (self-efficacy) and to make more effec-tive use of self -regulatory strategies like planning and self -monitoring (functioning self-regulation) as methods for achieving their goals. Hence, we posited that ado-lescents scoring higher on effortful control would be more confident about eating healthy and being physi-cally active under situations that may predispose them to eat unhealthy foods or be inactive. Furthermore, greater confidence (self-efficacy) would be related to healthier behavior in terms of better dietary intake patterns and more frequent physical activity. We also reasoned from a dual-process perspective that teenagers with higher scores on dysregulation will be more influenced by immediate situational pressures for potential rewards (e.g., fast food) and less inclined to invest effort in activities with longer payoff (e.g ., vigorous physical activity). thus, we pre-dicted that these youth would have unhealthier lifestyle behaviors, but this would occur more as a direct effect on behavior without necessarily involving self -efficacy measures that are specific to healthy eating and physical activity.

Methods

Participants

the study enrolled 1809 participants in grades 7–10 from 11 public schools in the bronx, New york, from an area that is characterized by relatively low income. inclu-sion criteria included adequate facility in english and not being enrolled in special education classes. thirty-eight participants were excluded from the analyses because they did not complete most of questionnaire. thirty per-cent of students were in 7th grade, 26% in 8th grade, 24% in 9th grade, and 20% in 10th grade. the 11 participating schools served a total of 11,789 students K–12, and 2,707 students were in grades 7–10. this cross-sectional study took place during the months of February–June and Octo-ber–December of 2008. the mean age of participants was 13.9 years (standard deviation [sD] 1.4, range 11–18) and 51% were female. the sample was 75% Hispanic, 6% African American, and 19% other race/ethnicity; 81% of participants were Us born. twenty-two percent of adoles-cents were overweight (85th–94th bMi percentile for age and sex) and 23% were obese (≥95th bMi percentile for age and sex). Data on maternal education indicated that 4% of mothers had only completed grade school, 19% had some high school education, 23% had graduated from high school, 15% had some college education, 28% were college graduates, and 8% had obtained a higher-level (master's, doctoral, or other professional) degree.

Procedure

the study was approved by the institutional Review boards of the Albert einstein College of Medicine and the New york City Department of education. before the start of the data collection, letters were mailed to par-ents informing them of the research study. in particular, parents were told to mail back a self-addressed prepaid postcard or to notify a designated school official if they did not want their child to participate in the study. Prior to data collection, students had the choice to refuse partici-pation or to sign an assent form if they agreed to partici-pate. the overall participation rate was 68% (1809/2707). Reasons for nonparticipation were parental or student refusals (1% and 9%, respectively), letters to parents returned by postal office (2%), and student absenteeism (20%); this absenteeism rate is consistent with typical absenteeism reported for schools in the area.40 A self-reported questionnaire was administered to students in classrooms by trained research staff. staff members used

a standardized script to provide group instructions to stu-dents and then addressed individual questions according to a defined protocol. trained research staff measured stu-dents' height and weight using a standardized procedure within one week of survey administration.

Measures

Demographics. the questionnaire included questions about participants' age, gender, and race/ethnicity, using the Us Census classification system for reporting race/ ethnicity. Parental education included items for father and mother and consisted of a 6-point scale with anchor points grade school and post-college education. because a large proportion of students did not know their father's education (64%), we used only maternal education in analyses.

Self-regulation measures. A multiple-indicator mea-surement model was used to assess self- control con-structs, similar to an approach used in prior studies.30,41,42 Responses for all items were on 5-point scales that ranged from “not at all true” to “very true.” effortful control was assessed by five indicators: a 6-item scale on planfulness (e.g., “i like to plan things ahead of time,” α = 0.70),43 a 6-item scale on problem solving (“When i have a prob-lem, i think about the choices before acting,” α = 0.87),43

a 5-item scale on soothability (“i can deliberately calm down when i am excited or wound up,” α = 0.76),43 a 7-item scale on delay of gratification (“ i can say No to a good time when i know there is work i have to do first,” α = 0.76),44 and a 7-item scale on self-reinforcement.45 Dysregulation was assessed by six indicators: a 6-item scale on impulsivity (e.g., “i often get in trouble because i do things without thinking,” α = 0.84),46 a 5-item scale on impatience (“i have to have everything right away,” α = 0.63),46,47 a 6-item scale on immediate gratification (“i have difficulty saving money to buy something several weeks later,” α = 0.78),44 a 6-item scale on angerability (“When i have a problem, i blame and criticize other peo-ple,” α = 0.81),48 a 5-item scale on distractibility (“i have to be reminded several times to do things,” α = 0.78),43 and a 6-item scale on self-criticism (“i seem to blame myself when things go wrong,” α = 0.79).45

Self-efficacy for healthy food choices. We used a 9-item scale validated by Neumark, et al.,49 which assesses self-efficacy to make healthy food choices in social, emotional, and normal situations (e.g., “if you wanted to, how sure are you that you could eat healthy food when you are… at a fast food restaurant”). Responses were on 5-point scales that range from “not at all sure” to very “sure.” the Cron-bach alpha for the study sample was 0.87.

Self-efficacy for being physically active. We used a 6-item scale validated by sallis, et al.50 the scale asked students how sure they were that they could do physi-cal activity in a variety of situations, such as when feel-ing sad or stressed, or when having a lot of homework. Responses were on 5-point scales that ranged from “i'm sure i can't” to “i'm sure i can.” the Cronbach alpha for the study sample was 0.85.

Dietary assessment. We assessed fruit and vegetable intake and snack and junk foods with the short version of the youth/Adolescent Questionnaire (yAQ), a food frequency questionnaire that was validated in the Grow-ing Up today study, and has shown an adequate cor-relation against the full yAQ and against 24-hr recall interviews.51,52 these food groups were chosen a priori because they are relevant to obesity risk. students were asked to report how often on average during the past year they had eaten the food items listed (e.g., apples, car-rots). Responses were on 8-point scales that ranged from “Never/less than 1 a month” to “More than 3 per day.” scores were computed by adding the individual responses to each food item within these food groups.

Physical activity. to assess physical activity we used questions from the youth Risk behavior survey.53 One item for vigorous exercise asked: “During the last 14 days, how many days have you done at least 20 minutes of exercise hard enough to make you breathe heavily and make your heart beat fast.” Responses for this item were on a 5-point scale (none, 1–2 days, 3–5 days, 6–8 days, 9 or more days).

Sedentary behavior. We used a sedentary behavior scale from Norman, et al.,54 which ask about how much time the participant spent doing sedentary activities the most recent day they were not in school. items include time spent watching television, playing computer or video games, sitting and listening to music, and sitting and talking on the phone or cell phone or text messaging, sedentary activities that are common among adolescents. in analyses, we excluded the item about time spent doing homework because school administrators informed us that homework was rarely assigned, and, consequently, few students reported spending any time doing home-work. Responses were on 9-point scales, from none to 6 hr or more (a = 0.68).

Height and weight. Height and weight were measured without shoes and with students wearing light clothes. Height was measured using a portable stadiometer (seca Portable stadiometer 214). Weight was measured using a digital portable scale (seca Robusta 813). these measures were used to derive a bMi z -score for each participant, using the CDC growth curves for age and sex.55

Statistical Analysis

Frequencies, means, and standard deviations were calculated for each variable. bivariate analyses included t-tests for differences in means between groups and zero-order correlation coefficients for continuous variables. For statistical analysis, we used a latent variable approach in structural equation modeling to test the relation of self-regulation constructs to diet, physical activity, and sedentary behavior and test for indirect effects through measures of self-efficacy for healthy eating and physical exercise. First, we conducted principal component analy-sis (PCA) to determine the presence of latent constructs. PCA showed higher loadings of effortful control indica-

tors in one factor, and higher loadings of dysregulation on a second factor, according to our hypothesized measure-ment model.

then, we conducted confirmatory factor analysis (CFA), using maximum likelihood estimation, to test the measure-ment structure and fit of the model. this CFA supported the hypothesized two-factor measurement model for self-reg-ulation constructs. We compared the fit of the two-factor model to alternative one-factor and three-factor models, specifying all 11 observed variables as indicators of one or three latent constructs of self- regulation. Nested tests showed that the two-factor model had better fit to the data than the alternative models (one- vs. two-factor model: dif-ference χ2 = 1795.34, degrees of freedom [df] = 1, p < 0.001; two- vs. three-factor model: difference χ2 = 416.34, df = 2, p < 0.001). self-regulation constructs specified as latent variables were indexed by five indicators for effortful control and six indicators for dysregulation. standardized loadings of indicators on constructs were between 0.51 and 0.82. the correlation of effortful control and dysregulation was −0.34.

to test the direct and indirect effects of self-regulation in relation to measures of dietary intake and physical activity, a structural equation model was specified having effortful control and dysregulation as exogenous vari-ables, adjusting for gender, ethnicity, and maternal edu-cation. Measures of self-efficacy for healthy eating and being physically active were specified as mediators, with a covariance of their residual terms. the criterion vari-ables in the model were fruit/vegetable intake, snack/junk food intake, physical activity, and sedentary behavior. All of these variables were specified with covariances of their residual terms. the structural model was analyzed using maximum likelihood methods and the expectation maxi-mization (eM) algorithm for missing data. standardized coefficients were then reported.

We also examined effect modification by gender and age group (11–13 vs. ≥14 years old). For these purposes, multiple group analyses were performed using structural equation models (seM) (gender and age group differ-ences were tested separately). subgroups were analyzed simultaneously with all parameters freely estimated. then, equality constraints were imposed to the paths from self-regulation constructs to behaviors (direct effect), from self-regulation constructs to self-efficacy measures (mediators), and from self -efficacy measures to behav-iors. Nested tests examined whether differences in coeffi-cients across groups were significant or not.56 Descriptive analyses were conducted with stata version 11, while confirmatory analyses and structural equation models were conducted with MPlus version 5.

Results

Descriptive Analysis

Descriptive statistics for self-regulation indicators and self-efficacy scales (table 1) showed that the variables

Table 1.

Descriptives for Continuous Variables (N = 1771)

Variable	Range	Mean	SD	Skewness	Kurtosis
Indicators of good self-control
Planfulness	6–30	21.4	4.4	–0.36	2.8
Problem solving	6–30	20.0	5.4	–0.19	2.7
Good delay of gratification	7–35	21.3	5.9	0.15	2.5
Soothability	5–25	14.1	4.6	0.17	2.5
Positive self-reinforcement	7–35	22.0	6.1	0.06	2.6
Indicators of poor regulation
Impulsivity	6–30	14.8	5.5	0.50	2.8
Poor delay of gratification	6–30	15.2	5.4	0.51	2.8
Angerability	6–30	13.1	4.9	0.95	3.6
Impatience	5–25	11.0	3.7	0.65	3.2
Distractability	5–25	11.6	4.4	0.64	3.0
Negative self-reinforcement	6–30	14.1	4.9	0.64	3.3
Food self-efficacy	9–45	23.8	8.3	0.27	2.7
Physical activity self-efficacy	6–30	18.0	5.9	–0.02	2.5
Fruit/vegetable	4–32	14.5	6.8	0.71	2.8
Junk food/snacks	9–64	30.6	12.3	0.66	2.8
Vigorous activity	1–5	2.8	1.3	0.32	2.0
Sedentary behavior	5-45	23.8	7.6	–0.39	2.9

had relatively normal distributions. table 2 shows differ-ences in self-efficacy measures and lifestyle behaviors by demographic characteristics. self-efficacy for healthy eating and being physically active were lower in Hispan-ics (23.5 vs. 24.8 p < 0.05 and 17.8 vs. 18.6, p < 0.05, respectively) and in girls (23.5 vs. 24.3 p < 0.05 and 16.6 vs. 19.7, p < 0.001, respectively). in addition, His-panic teens showed lower fruit/vegetable intake and lower physical activity, compared to non-Hispanic adolescents (14.2 vs. 15.2, p < 0.05 and 2.7 vs. 2.9, p < 0.05, respec-tively). Compared to boys, girls reported higher intake of junk food/snacks (31.9 vs. 29.2, p < 0.001), more time in sedentary behaviors (24.6 vs. 22.8, p < 0.001), and lower

Table 2.

Mean Distribution of Self-Efficacy Measures and Lifestyle Behaviors (N = 1771)

					Maternal	Maternal
					education less	education high
	Girls	Boys	Hispanic	Non-Hispanic	than high school	school/higher
	Mean (SD)	Mean (SD)	Mean (SD)	Mean (SD)	Mean (SD)	Mean (SD)
Food self-efficacy	23.5 (8.5)*	24.3 (8.0)	23.5 (8.3)*	24.7 (8.2)	22.6 (8.5)*	23.9 (8.1)
PA self-efficacy	16.7 (5.9)***	19.7 (5.8)	17.8 (6.1)*	18.6 (5.6)	18.1 (6.1)	17.9 (5.9)
Fruit/vegetable	14.5 (7.1)	14.5 (6.7)	14.2 (6.9)*	15.3 (6.7)	14.1 (6.9)	14.5 (6.8)
Snacks/junk food	31.9 (12.7)***	29.2 (11.7)	30.4 (12.3)	31.0 (12.4)	30.8 (12.3)	30.6 (12.3)
Vigorous activity	2.4 (1.2)***	3.2 (1.3)	2.7 (1.32)*	2.9 (1.3)	2.6 (1.3)	2.79 (1.3)
Sedentary behavior	24.6 (7.8)***	22.8 (7.4)	23.7 (7.7)	24.1 (7.6)	24.2 (7.6)	23.7 (7.6)

^*p < 0.05. ^***p < 0.001.

 SD, standard deviation.

levels of physical activity (2.3 vs. 3.2, p < 0.001). Cor-relations among independent and criterion variables are presented in table 3. intercorrelations of the criterion variables ranged from 0.05 to 0.39.

Table 3.

Correlations between Study Variables (N = 1771)

	1	2	3	4	5	6	7	8
1. Good self-control	—
2. Poor regulation	–0.30***	—
3. Food self-efficacy	0.26***	–0.01	—
4. PA self-efficacy	0.29***	–°.09**	0.23***	—
5. Fruit/vegetable	0.15***	0.05*	0.17***	0.08**	—
6. Snacks/junk food	–0.13***	0.31***	–0.02	–0.07**	0.37***	—
7. Vigorous activity	0.16***	–0.09**	0.12***	0.39***	0.10**	–0.06*	—
8. Sedentary behavior	–0.10**	0.33***	–0.02	–0.04	0.16***	0.36***	–0.05*	—
9. Age	–0.05	0.02	–0.02	0.07**	–0.03	0.05	–0.02	0.06*

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

Structural Equation Modeling Analysis

the structural equation model, specified as described in the methods section, showed a good fit to the data (chi-squared = 685.7, df = 147; N = 1771; comparative fit index [CFi] = 0.93; root mean square error of approximation [RMseA] = 0.045). Results (Fig. 1) showed that effortful control had significant positive path to both measures of self-efficacy, and there was a path from self-efficacy for

Figure 1.

Structural equation model for the association of self-regulation constructs and lifestyle behaviors (N = 1771). Chi-squared test of model fit = 685.7, degrees of freedom (df ) = 147; comparative fit index (CFI) = 0.93; root mean square error of approximation (RMSEA) = 0.045. Results are controlled for age, gender, ethnicity, and maternal education. Values represent standardized coefficient. Depicted paths are significant at p < 0.05. Problem solv., Problem solving; Pos. Self. Reinf., positive self-reinforcement; Neg. Self. Reinf., negative self-reinforcement.

healthy eating to fruit/vegetable intake, producing a signifi-cant indirect effect (p = 0.001). effortful control also had a direct effect on higher fruit/vegetable intake, suggesting that the relation of effortful control with fruit/vegetable intake was only partially mediated by self-efficacy. effort-ful control had significant indirect effects on higher vigor-ous physical activity (p < 0.001) through self-efficacy for being physically active with no direct effect, which indi-cates that the association of effortful control with physical

activity was fully mediated by self-efficacy. Dysregulation had a strong direct effect to higher intake of junk food/ snacks and to time spent in sedentary behavior and a small inverse path to vigorous activity level. but dysregulation did not show an association with self-efficacy, and there-fore, the self-efficacy measures included here did not show mediation for the association of dysregulation with lifestyle behaviors. All of the effects reported here were controlled for covariance and path effects for demographic character-

istics, but these coefficients are excluded from the figure for graphical simplicity.

snack/junk food intake and sedentary behaviors sig-nificantly increased with age (β = 0.07, p < 0.001 and β

 = 0.09, p < 0.01, respectively), while physical activity decreased (β = −0.06, p < 0.01). Compared to adolescent boys, girls had higher intake of snack/junk food (β = 0.12, p < 0.001) and lower physical activity (β = −0.23, p < 0.001). Race/ethnicity and maternal education were not significantly associated with criterion variables. self-efficacy for healthy eating was lower among adolescents whose mother did not graduate from high school (β = −0.06, p < 0.05); self-efficacy for being physically active was lower among girls (β = −0.26, p < 0.001). Age was significantly correlated with self-efficacy for being physi-cally active (β = 0.09, p < 0.01) but not with self-efficacy for healthy eating. Race/ethnicity was not associated with either measure of self-efficacy. Girls and Hispanics had lower scores on effortful control (β = −0.07, p < 0.01 and β = −0.06, p < 0.05, respectively), but dysregulation was not associated with any demographic characteristic in this sample. A second analysis was conducted including the bMi z-score as criterion variable. in this model, there were no associations of self-regulation or self-efficacy constructs with bMi z-score (data not shown).

We also conducted multiple group structural equation models to examine gender and age differences. the find-ings reported here did not differ by gender. However, the association of effortful control with self-efficacy was significantly stronger in the older group (≥14 years) than in the younger group (11–13 years old)(β = 0.31 vs. β = 0.23, p < 0.05). in addition, there was an inverse asso-ciation of dysregulation with physical activity among younger adolescents (11–13 years) (β = −0.16, p < 0.01) that was not observed in the older group (β = −0.01).

Discussion

the aim of this study was to test a theoretical model of self-regulation constructs and weight- related behaviors in a sample of inner-city adolescents. As predicted, the results indicated a direct effect of effortful control on higher fruit/vegetable intake as well as indirect effects on several health-promoting behaviors through self- efficacy for healthy eating and being physically active. Also as predicted, dysregulation was related to unhealthier behav-ior patterns, including higher junk food/snack intake and time spent in sedentary behavior, but these relations occurred as direct effects.

the present results converge with a previous study of adolescents30 in that the two domains of self-regulation showed different patterns of relations to diet and exercise behaviors. this is consistent with dual-process models of health behavior, suggesting the operation of two types of processes in shaping health-related behaviors.16,36 the study findings are also consistent with Hoffman, et al., who showed that candy intake was associated with

impulsivity.57 the variables involved in the construct for effortful control represent what is termed the reasoned or controlled pathway, whereas the indicators of dys-regulation are consistent with what is termed a reactive or impulsive pathway.36,58 A new aspect of the present study was the inclusion of self-efficacy measures to examine how self-control fits within the social cognitive theory framework. the findings suggest that greater effortful control is related to more confidence for engaging in healthy behaviors. being able to employ self-regulatory strategies such as problem solving could translate into greater confidence of accomplishing behavioral goals. this is relevant for several areas because studies have shown that more efficacious adolescents are more physi-cally active.59,60 self-efficacy is also related to dietary behaviors, including higher intake of fruit and vegeta-bles,61 choosing low-fat snacks,62 and less consumption of foods high in fat and sugar content.63 in this sample, dys-regulation was not associated with self-efficacy. this lack of association could be explained by the reactive nature of dysregulation processes–the tendency for acting impul-sively and seeking immediate rewards. it should also be noted that the measures of self-efficacy used in the study were specific for engaging in healthy lifestyle behaviors. it is possible that dysregulated adolescents are more sus-ceptible to temptations for eating snacks or junk food or watching tV. because we did not measure this type of temptation resistance, we cannot rule out that dysregula-tion is inversely related to other aspects of self-efficacy.

the latent constructs of self-regulation used in this research, which are broader than the single indices used in some previous studies, suggest a somewhat differ-ent perspective on weight- related behaviors. the way in which aspects of effortful control such as planning and problem solving relate to healthy dietary intake raises questions about how these are involved in gathering information and making decisions about eating, while the element of soothability addresses an emotional compo-nent in health maintenance that may not be distinct from other aspects of self-regulation.48 self-reward can also be crucial for healthful diet patterns through reinforcing difficult choices and reinforcing progress toward self-defined goals. Analogously, the dysregulation construct includes not just impulsiveness but also elements such as self-criticism and poor control of anger. the fact that these elements also relate to unhealthy behavior patterns such as junk food intake and sedentary activity brings up questions about whether such behaviors are engaged in simply because they do not require effort or because other variables ( e.g., media advertising, susceptibility to peer influences) are also relevant. We think that through raising such questions the self-regulation perspective sug-gests a detailed approach to further investigations using epidemiologic designs.

in this sample, we found lower effortful control among Hispanic adolescents when compared to non-Hispanic youth. Data about ethnic differences in self-regulation

characteristics are scarce, so this finding needs replica-tion. Furthermore, our predominantly Hispanic sample prevented us from examining differences across other racial/ethnic groups. the study did not have extensive measures for socioeconomic status and familial or envi-ronmental factors that could help us explain the ethnic differences observed in the study. A previous study showed that young children with low self-regulatory skills had mothers with lower educational attainment and lived in families with lower family income.20 On the other hand, Hispanics in the United states have lower educa-tion levels and are more likely to live in poverty than other non-Hispanic groups.64,65 thus, it is possible that the observed differences are largely due to differences in socioeconomic status rather than ethnicity.

the present study has some limitations. the study had

a cross-sectional design, therefore temporal relationships cannot be definitively established. Due to the constraints to fit the classroom assessments within one class period, the study had to rely on a brief instrument for dietary assessment and in a simple self-reported measure of physical activity. Furthermore, the study did not include independent ratings (e.g., teachers, parental) for measur-ing self-regulation constructs. these limitations may have introduced measurement error. However, this type of mea-surement error is nondifferential and may have attenuated the magnitude of the associations observed. the model did not include measures of parental influences and environ-mental variables, which research has shown to be associ-ated with children's diet and physical activity. Finally, bMi in early adolescence was not related to self-control measures in the present study. this may be attributable to the relatively high rate of obesity in the present sample, or; perhaps this association develops over considerable period of time, and is better observed in long-term studies.

the present study included only a couple of indicators of emotion regulation (soothability and angerability), a dimension of self-regulation that is relevant in effortful control and dysregulation.35,48,66,67 Prior studies examining emotional eating suggest that emotion regulation plays a role on eating behaviors and obesity.68,69 Furthermore, Graziano, et al. showed that poorer emotional regulation predicted excess weight among toddlers.70 thus, future studies need to address this aspect of self-regulation for a better understanding of the effects of effortful control and dysregulation on obesity risk.

Conclusions

Despite the study limitations, the present findings obtained from a sample of inner-city adolescents contribute to the growing literature on self- regulation and weight-related behaviors. self-regulation skills can be taught, as shown in several studies conducted in educational set-tings71–73 and other studies related to obesity prevention.74,75 Furthemore, efforts to improve effortful control skills could be a good strategy for improving resilience skills in minor-

ity and economically disadvantaged children and in this way help to counteract deleterious effects of the stressful and obesogenic environments in which they grow up.

Author Disclosure Statement

No competing financial interests exist.