Research ArticleRacial and Ethnic Differences in the Home FoodEnvironment Explain Disparities in Dietary Practicesof Middle School Children in TexasNalini Ranjit, PhD; Alexandra E. Evans, PhD; Andrew E. Springer, DrPH;Deanna M. Hoelscher, PhD; Steve H. Kelder, PhD

MichaelCenter aAddressHealthyHealth, A(512) 39�2015 Shttp://dx

Journal

ABSTRACT

Objective: To examine racial and ethnic differences among middle school children in the home foodenvironment (HFE) and the extent to which associations of healthy and unhealthy eating with the HFEdiffer by race and ethnicity.Design: Cross-sectional secondary analyses of baseline data from Coordinated Approach to Child HealthMiddle School, a school-based intervention targeting obesity and obesogenic behaviors among middleschool children in Austin, TX.Participants: A total of 2,502 children (mean age, 13.9 years; 58%Hispanic, 28%white, and 14% black).Variables Measured: Availability and accessibility of healthy foods, and parental support of healthyeating, and family meals. Consumption of both healthy and unhealthy foods was examined.Analysis: Differences across racial and ethnic groups in aspects of HFE were estimated using linear regres-sion. Models also examined racial and ethnic differences in consumption of healthy and unhealthy foods. Ifadjusting for HFE, such differences were accounted for.Results: White children had significantly better HFEs than Hispanic and black children with greateravailability and accessibility of healthy foods (P < .001). Adjusting for a healthy HFE reduced disparitiesin consumption of healthy foods but not in consumption of unhealthy foods.Conclusions and Implications: Improved HFE may increase healthy eating among ethnic minoritiesbut is unlikely to reduce unhealthy eating.Key Words: home food environment, racial/ethnic disparities, healthy eating, unhealthy foods (J NutrEduc Behav. 2015;47:53-60.)

Accepted September 7, 2014. Published online October 25, 2014.

INTRODUCTION

Disparities in childhood obesity byrace and ethnicity have continued toexpand over time while the overallpace of obesity increase has begunto slow down.1-3 Such differences inobesity emerge as early as preschool,4

continue into adulthood, and mayeven widen over the life course.5

These pernicious disparities in obe-sity reflect the influence of both die-tary and physical activity behaviors,which have been shown in multiple

and Susan Dell Center for Healthy Lt Houston School of Public Health, Aufor correspondence: Nalini Ranjit, PLiving, University of Texas Health Sustin Regional Campus, 1616 Guadalu

1-2527; Fax: (512) 482-6185; E-mail: NOCIETY FOR NUTRITION EDUC.doi.org/10.1016/j.jneb.2014.09.001

of Nutrition Education and Behav

studies to be patterned by race andethnicity.6-8 There is a general con-sensus that this patterning of behaviorstems from the constellation of ec-onomic, social, and cultural factors thatcharacterize the usual living conditionsof poor and disadvantaged minorities.For example, extensive research do-cuments that the communities inwhich ethnic minorities reside aredisproportionately subject to increasedavailability of energy-dense foods,reduced availability of affordablehealthy options, and built environ-

iving, University of Texas Health Sciencestin Regional Campus, Austin, TXhD, Michael and Susan Dell Center forcience Center at Houston School of Publicpe St, Ste 6.300, Austin, TX 78701; Phone:alini.Ranjit@uth.tmc.eduATION AND BEHAVIOR

ior � Volume 47, Number 1, 2015

ments that offer little opportunity forsafe recreational physical activity.9-11

However, there is little literaturedocumenting whether exposure tomore proximal environments, such asthe home food and activity envi-ronments, is similarly patterned, and ifso, whether such differential exposureunderlies the observed racial andethnic disparities in obesity and inbehaviors leading to obesity.11

Substantial evidence points to thehome food environment, as structuredby parents, as a key cause of poor die-tary practices and obesity. Parentsmodel eating behaviors, food choices,and taste preferences12,13; in addition,they can employ specific behavioralstrategies such as restricting orencouraging consumption of specificfoods and food groups14 and structurethe home food and mealtime environ-ments to facilitate availability andaccessibility of specific foods15,16 aswell as regular family meals together,

53

54 Ranjit et al Journal of Nutrition Education and Behavior � Volume 47, Number 1, 2015

including breakfast at home.17,18

Although much of this literature isfocused on the development of youngchildren's eating habits,19-21 researchsuggests that familial influenceson eating habits continue into ado-lescence.15,22,23

Surprisingly, given the importanceof the home food environment, thereis little research examining the preva-lence and consequences of systematicdifferences in the home food environ-ment by race and ethnicity, althoughit is known that such environmentsare socioeconomically patterned.16

In the research reported here, theauthors sought to address this gap inthe literature by examining differ-ences in home food environments ofmiddle school students by race andethnicity and by exploring the roleof home food environment in dietand obesity disparities. Specifically,the objectives of this study were toexamine (1) the magnitude of racialand ethnic gaps on multiple measuresof the home food environment, (2)differences in healthy and unhealthydiet and eating practices across racialgroups, and (3) the extent to whichracial and ethnic differences in dietand eating practices are explainedby differences in the home food envi-ronment.

METHODSData and Measures

Data for these analyses were drawnfrom the baseline sample of the Co-ordinated Approach to Child Health(CATCH) Middle School program, aschool-based obesity intervention pro-gram designed for students in the Cen-tral Texas area.24 Eighth-grade studentsfrom 30 middle schools randomlyselected from 5 school districts in thearea were invited to participate, andmeasures were administered to con-senting students in 2009. Measuresrelevant to this analysis were primarilyself-reported measures, collected viasurveys completed by students in theclassroom during advisory period. Inaddition, trained study staff measuredheight with a portable stadiometer(Portable Adult Measuring Unit PE-AIM-101, Perspective Enterprises, Por-tage, MI) and weight with a portabledigital scale (Tanita BWB-800S, TanitaCorporation of America, Inc, Arlington

Heights, IL). All analyses reportedhere are cross-sectional and used onlythe baseline data from this program,collected in 2009.

Demographic data available forstudents included age in years, sex,and ethnicity. These analyses werelimited to children from the 3 majorethnic groups: white, black, andHispanic. School-level data indicatingthe percentage of children eligible forfree and reduced meals obtained fromstate administrative records were usedas a proxy measure of school-leveleconomic disadvantage or socioeco-nomic status (SES), following previousresearch with Texas students.25,26

Body mass index (BMI) was obtainedfrom height and weight data andconverted to age- and sex-specificBMI z-scores, as per guidelinesfrom the Centers for Disease Controland Prevention.27 Language spokenat home (English or other) was usedas a measure of acculturation.

Measures of the home food envi-ronment obtained in the study wereadapted from published studies28,29

and have adequate face and contentvalidity. These analyses included mea-sures of home availability of healthyfoods in the past week (range, 0–9;obtained as a sum of 3 items [avai-lability of vegetables, fruit, and fruitjuice], each scored on a 4-point ratingscale ranging from never, throughsome of the time, most of the time, andall the time), and easy access to healthyfoods in the past week (range, 0–6; ob-tained as a sum of 2 items [fruitsand vegetables], each with 4 ordinalresponse categories as listed above).Although related, availability andaccessibility measure different aspectsof the home food environment; spe-cifically, both availability and ac-cessibility are needed to motivatechildren's consumption of fruits andvegetables when preference for theseitems is low.30 Other items includedparental encouragement of healthyeating (range, 0–20; which included5 Likert-scaled items pertaining toconsumption of fruits and vegetables,low-fat milk, water instead of a sugar-sweetened beverage (SSB), wholegrain, and regular breakfast; responseoptions were in 5 graded categoriesfrom never to always), number of fam-ily meals in the past week (range, 0 to$ 7), and number of days that break-fast was consumed at home in the

past week (range, 0–7). A compositehome food environment score wasconstructed as the sum of all of theseitems after rescaling them to SD units(range, –10.1 to 6.5; mean, 0 [SD, 3.1];Cronbach a ¼ .66).

Measures of dietary practices weredrawn from 11 items on the surveythat queried respondents on thefrequency of usual weekday consump-tion of specific marker foods or foodgroups. Response categories on theseitems ranged from never through $ 5times/d. A healthy diet index (range,0–20; Cronbach a ¼ .66) was devel-oped for this study as a sum of itemsreferencing usual consumption offruit (including juice) and vegetable,milk, and whole grains. Likewise, anunhealthy diet index (range, 0–15;Cronbach a ¼ .69) was constructedas a sum of items examining usualconsumption of chips, SSBs (sodaand fruit-flavored drinks), and caffein-ated beverages (coffee or tea andenergy drinks). In addition, some ofthe constituent items were examinedsingly or in groups as follows: usualfruit and vegetable intake, includingjuice (range, 0–15), usual healthy bev-erages (milk or water; range, 0–10),usual SSB intake (range, 0–10), andusual salty snack intake (range, 0–5).Another dietary constituent of inter-est to the researchers was dessertintake. Because there were no ques-tions relating to usual intake of sweetsand dessert, reported frequencies ofprevious day consumption of each of4 different types of dessert itemswere summed into a dessert intakeindex (range, 0–20; Cronbach a ¼.84), which was then used as a proxyfor usual dessert intake.

Statistical Analysis

Socioeconomic and demographic cha-racteristics of the data were examinedfor the total sample, as well as for eachof the 3 racial and ethnic groups, asmeans for continuous measures, andfrequency distributions for categoricalmeasures. The researchers evaluateddifferences across racial and ethnicgroups on these characteristics usingF test for continuous measures andchi-square statistics for categoricalmeasures (Table 1). Spearman's corre-lations, partialing out the effects ofage, sex, and BMI, between each

Table 1. Demographic Characteristics of Study Sample, by Race and Ethnicity, CATCH Middle School Project, Spring, 2009

Total Black Hispanic WhiteDifference AcrossCategories (P)

Number of subjects, n (%) 2,502 365 (14.6) 1,440 (57.6) 697 (27.9)

GenderMale, n (%) 1,204 (48.2) 55.9 46.3 48.1 .005

Mean age, y (mean [SD]) 13.88 (0.59) 13.9 13.9 13.8 < .001

Weight status categoryNormal, n (%) 1,464 (61.3) 59.7 55.1 75.3Overweight, n (%) 457 (19.2) 19.3 21.7 13.7Obese, n (%) 466 (19.5) 21.0 23.2 11.0 < .001

Language spoken at homeEnglish, n (%) 1,657 (67.2) 97.8 44.1 98.4Other, n (%) 808 (32.8) 2.2 55.9 1.6 < .001

Socioeconomic status (based onschool-level measure)

Percent eligible for free or reducedmeals (mean [SD])

60.9 (28.2) 65.9 71.4 36.5 < .001

CATCH indicates Coordinated Approach to Child Health.Note: Tests of significance for categorical variables are based on chi-square tests and examine whether the distribution of de-mographic characteristics is comparable across race groups. The F test was used to compare the distributions of continuousmeasures, including age and socioeconomic status.

Journal of Nutrition Education and Behavior � Volume 47, Number 1, 2015 Ranjit et al 55

measure of the home food environ-ment and usual healthy or unhealthydiet were examined to confirmwhether associations in these datawere in line with the general findingsin the literature (Table 2). Correlationswere obtained for the entire sample aswell as for each racial and ethnic group.Analyses presented inTable 3 sought toexamine whether there were system-atic differences in measures of the

Table 2. Correlations Between Home FoodSpring, 2009 (n ¼ 2,826)

Home EnvironmentMeasure

HealthyDietScore

Home availability of healthyfoods

0.25

Easy accessibility of healthyfoods

0.23

Parental support of healthy diet 0.29

Number of family meals in pastweek

0.18

Number of breakfasts at homein past week

0.13

Sum of healthy homeenvironment measures1

0.32

CATCH indicates Coordinated ApproachNote: Pairwise Spearman’s correlations, pz-scores, were estimated. Only statistical

home food environment across racialand ethnic groups, using estimatesfrom linear regressions adjusting for so-ciodemographic measures includingage, sex, individual SES, and school-level SES. In the final set of analyses(Tables 4 and 5), dietary practices wereexamined using linear regressionmodels, with racial and ethnicdifferences in these practices obtainedby contrasts. Three nested models

Environment Measures and Dietary Intake Me

Fruitsand

VegetablesHealthy

BeveragesUnhealthyDiet Score

0.32 0.17 �0.19

0.27 0.16 �0.13

0.25 0.21 �0.23

0.18 0.13 NS

0.09 0.11 �0.15

0.34 0.24 �0.22

to Child Health; NS, not significant.artialing out the effects of age, sex, languagely significant estimates (ie, P < .05) are repo

were examined: Model 1 obtainedracial and ethnic differentials afteradjusting for possible confounding byage, sex, acculturation, and BMIz-scores; Model 2 added the compositehome food environment measure tothe covariates in Model 1; and Model3 examined the impact of adding theSES measure (percent economicallydisadvantaged) to Model 2. The au-thors used P for differences in dietary

asures, CATCHMiddle School Project,

Sugar-SweetenedBeverages

SaltySnacks Desserts

�0.20 �0.13 �0.08

�0.14 �0.06 �0.07

�0.23 �0.14 �0.10

�0.05 NS NS

�0.19 �0.09 �0.05

�0.25 �0.14 �0.09

spoken at home, and body mass indexrted.

Table 3. Home Food Environment, by Race and Ethnicity, CATCH Middle School Project, Spring, 2009

Home Environment Measure (Range) White (n ¼ 697) Black (n ¼ 365) Hispanic (n ¼ 1,440)

Mean (SE) Mean (SE) P Mean (SE) P

Availability of healthy foods (0–9) 6.5 (0.1) 6.0 (0.12) < .001 5.9 (0.06) < .001

Easy accessibility of healthy foods (0–6) 4.1 (0.08) 3.8 (0.1) .006 3.8 (0.05) .01

Parental support of healthy diet (0–20) 13.3 (0.27) 12.5 (0.35) .05 12.7 (0.17) .07

Number of family meals in past week (ordinal 0–5) 2.5 (0.09) 2.2 (0.11) .08 2.5 (0.05) .92

Number of breakfasts at home in past week (ordinal 0–5) 3.1 (0.1) 2.0 (0.12) < .001 2.2 (0.06) < .001

Standardized sum of all home food environment foodmeasures (–10 to 6.6)

0.9 (0.15) -0.4 (0.19) < .001 -0.2 (0.09) < .001

CATCH indicates Coordinated Approach to Child Health.Note: Estimates are derived from regression models adjusted for age, sex, language spoken at home, and school socioeco-nomic position. Tests of significance, based on the F test, are reported using the value for white children as referent.

56 Ranjit et al Journal of Nutrition Education and Behavior � Volume 47, Number 1, 2015

practices across racial and ethnicgroups to evaluate the strength ofracial and ethnic differences in eachmodel. All hypotheses were 2-tailedand the threshold for significance wasset at .05. Analyses were completedusing SAS, version 9.3 (SAS Institute,Cary, NC, 2012).

Approval for the study CATCHMid-dle School Study was obtained fromthe University of Texas Health ScienceCenter at Houston Committee for theProtection of Human Subjects as wellas from participating school districts.

RESULTS

A total of 30 schools participated inthe study. The SES of these schools,as measured by the percentage ofchildren eligible for free or reduced-price lunch, varied widely across theschools (10.5% to 95.8%), but a me-

Table 4. Differences Between White andAdjusting for Socioeconomic Sta

Healthy diet score

Fruit and vegetables (includes fruit juice)

Healthy beverages (milk and water)

Unhealthy diet score

Sugar-sweetened beverages

Salty snacks

Desserts

CATCH indicates Coordinated Approach

dian value of 72% on this measuresuggests that these schools weregenerally composed of low-SES stu-dents. Of the 2,821 students whowere surveyed at baseline, 2,502 ofwhite, Hispanic, or black race orethnicity were retained for analysis.Over half of the students in the studysample (53%) were Hispanic, whereasa little under a quarter were white(24%) (Table 1). Mean age of His-panics was slightly higher than theother groups, whereas females wereoverrepresented among blacks, rela-tive to whites and Hispanics. Over athird of the sample, mostly Hispanicchildren, reported speaking in a lan-guage other than English at home.Blacks and Hispanics were over-represented in schools with highproportions (> 60%) of economicallydisadvantaged students. Over 40% ofblacks and Hispanics were overweight

Black Middle School Children in Consumptus and Home Food Environment, CATCH M

Model 1 Mode

Mean (SE) P Mean (SE)

0.24 (0.23) .31 0.89 (0.23)

0.63 (0.2) .002 1.21 (0.19)

0.02 (0.16) .90 0.35 (0.15)

2.92 (0.27) < .001 2.6 (0.27)

1.88 (0.15) < .001 1.68 (0.15)

0.68 (0.08) < .001 0.61 (0.08)

2.91 (0.27) < .001 2.74 (0.27)

to Child Health.

or obese, compared with 25% ofwhites. Overall, the table confirmsmarked differences in the demo-graphic and socioeconomic profilesof children from different ethnicgroups.

Association of Home FoodEnvironment With DietaryPractices

Table 2 confirms that a healthy foodenvironment is positively related todietary practices. Pairwise Spearman'scorrelation coefficients partialing outthe effects of age, sex, BMI z-score,and acculturation were obtained foreach combination of home food en-vironment measure with each dietarymeasure. Healthy food availabilityand parental support were signifi-cantly related to all of the dietarymeasures and in the expected

tion of Index Foods, Before and Afteriddle School Project, Spring, 2009

l 2 Model 3

P Mean (SE) P

< .001 0.66 (0.24) .005

< .001 1.07 (0.2) < .001

.02 0.35 (0.16) .03

< .001 2.04 (0.29) < .001

< .001 1.39 (0.15) < .001

< .001 0.51 (0.08) < .001

< .001 2.31 (0.28) < .001

Table 5. Differences Between White and Hispanic Middle School Children in Consumption of Index Foods, Before and AfterAdjusting for Socioeconomic Status and Home Food Environment, CATCH Middle School Project, Spring, 2009

Model 1 Model 2 Model 3

Mean (SE) P Mean (SE) P Mean (SE) P

Healthy diet score �0.16 (0.2) .40 0.31 (0.19) .10 0.10 (0.2) .61

Fruit and vegetables (includes fruit juice) 0.08 (0.17) .64 0.52 (0.16) .001 0.39 (0.17) .02

Healthy beverages (milk and water) �0.29 (0.13) .02 �0.06 (0.13) .68 �0.06 (0.14) .66

Unhealthy diet score 2.14 (0.23) < .001 1.88 (0.23) < .001 1.35 (0.24) < .001

Sugar-sweetened beverages 1.19 (0.12) < .001 1.03 (0.12) < .001 0.75 (0.13) < .001

Salty snacks 0.52 (0.07) < .001 0.48 (0.07) < .001 0.38 (0.07) < .001

Desserts 1.42 (0.22) < .001 1.32 (0.23) < .001 0.91 (0.24) .001

CATCH indicates Coordinated Approach to Child Health.Note: Model 1 adjusts for age, sex, language spoken at home, and bodymass index z-score; Model 2 is Model 1 plus summaryhome food environment score; Model 3 is Model 2 plus socioeconomic status. Positive values indicate a higher value on theoutcome measure for black or Hispanic children relative to white children. All tests of significance are based on the F test.

Journal of Nutrition Education and Behavior � Volume 47, Number 1, 2015 Ranjit et al 57

direction, although the relationshipwas strongest for fruits and vegetables(r ¼ 0.32 for availability and r ¼ 0.25for parental support), as also reflectedin the healthy diet index (r ¼ 0.25and r ¼ 0.29 for availability andparental support) (Table 2). Only thenumber of family meals in the pastweek did not show the expectedassociations in the total sample. Thecomposite measure of healthy foodavailability, presented in the last rowof the table, encapsulated theseindividual effects, and the resultsconfirmed that it was an adequatesummary measure of the home foodenvironment. In addition, these asso-ciations were examined separately byracial and ethnic groups (data notshown). Although the stratified ana-lyses were largely consistent with thefull sample analyses, some importantdifferences were noted. Overall, associ-ations of eachhome food environmentmeasure with both healthy and un-healthy diet outcomes were strongestfor white children and weakest forblack children in the sample. In partic-ular, among blacks, there was littleevidence of a negative association be-tween healthy home food environ-ment and unhealthy diet.

Racial and Ethnic Differences inHome Food Environment

Consistent racial andethnicdifferencesin different individual aspects of thehome food environment, as well as in

the overall home food environment,are evident in Table 3. White childrenreported significantly higher availa-bility (P < .001) and accessibility ofhealthy foods (P < .007), greaterparental support of healthy diet, andhigher frequency of consuming break-fast at home relative to black andHispanic children. Significant racialand ethnic differences in home foodenvironment (whites ¼ 0.9 vs blacks¼ –0.4, P < .001; whites ¼ 0.9 vsHispanics ¼ 0.2, P < .001) were re-flected in the composite measure,which captured the combined preva-lence of each of these aspects of thehome food environment. In mostcases, black and Hispanic children hadcomparable values for these measures.The large magnitude of racial andethnic differences on the compositemeasure suggests substantial clusteringof adverse home food environments.Indeed, correlations among the indi-vidual home food environmentmeasures (data not shown) rangedfrom 0.20 to 0.61.

Role ofDisparities inHome FoodEnvironment in ExplainingRacial andEthnicGaps inDietaryPractices

Table 4 examines the extent to whichracial and ethnic differences in theconsumption of healthy and un-healthy foods can be explained byhome food environment and socio-economic position. All differences

are presented so that positive valuesimply higher consumption amongthe index racial or ethnic group rela-tive to white children (the referentgroup). In general, unadjusted levelsof consumption of healthy foods,especially fruits and vegetables,among black middle school studentswere comparable to those amongtheir white counterparts (Model 1).After adjusting for home food envi-ronment (Model 2), the racial differ-ence expanded substantially in favorof blacks and became significant (esti-mated excess consumption amongblacks was 0.66, P ¼ .005, for healthydiet scores; and 1.07, P < .001, forfruits and vegetables). Thus, withsimilar home food environment,blacks consumed healthy foods morefrequently than whites. Similarly, thegap between Hispanic and whitestudents in consumption of fruitsand vegetables changed in favor ofHispanics, from a nonsignificant .08before adjustment to .39 (P ¼ .02)after adjusting for home food envi-ronment. Similar changes favoringHispanics were also seen with healthybeverage consumption, with a signifi-cant gap favoring whites contrastingwith a nonsignificant gap after adjust-ment. In both groups, home foodenvironment was a more powerfulpredictor of racial gaps in consump-tion of healthy foods than socioeco-nomic position.

Consumption of unhealthy foodswas considerably higher among blacksand Hispanics relative to whites.

58 Ranjit et al Journal of Nutrition Education and Behavior � Volume 47, Number 1, 2015

Adjusting for home food environmentand SES reduced the gap, particularlyfor SSB (adjusting for home food envi-ronment reduced the gap from 1.88 to1.39 for blacks, and from 1.19 to 0.75for Hispanics), but differences acrossrace and ethnicity continued to remainlarge and significant.

DISCUSSION

These cross-sectional analyses exam-ining racial and ethnic differencesamong various aspects of the homefood environment in a multiethnicsample of middle school childrenprovide several insights into racialand ethnic disparities in diet. As hasbeen shown elsewhere,31 white chil-dren have a small but consistentadvantage over black and Hispanicchildren on several individual mea-sures of the home food environment,including availability and accessibilityof healthy foods, and breakfast athome, as well as a substantial advan-tage on a cumulative measure of thehome food environment. Positive asso-ciations between the home food envi-ronment and healthy diet existamong all racial and ethnic groups butare strongest among white children.This pattern of differences across racialand ethnic groups was even moreevident in the associations betweenhome food environment measuresand unhealthy diet and eating as well,with strong negative associationsobserved for whites and relativelyweaker associations among Hispanicand black children.

Novel findings emerged in thisexamination of racial and ethnicdifferences in diet once the re-searchers took into account variationsin home food environment. The gapbetween white children and childrenof minority groups in consumptionof healthy foods (fruits, vegetables,milk, water, and whole grains) is smalland may even favor the minoritygroups. Indeed, the results suggestthat black and Hispanic childrenexposed to healthy food environ-ments would consume more healthyfoods than would white children. Incontrast, consumption of unhealthyfoods such as SSBs, salty snacks, anddesserts is consistently higher amongracial and ethnic minorities and notsufficiently explained by differences

in the home food environment. It ispossible that parents who are con-strained in their ability to ensure ahealthy home food environment arelikewise limited in their food choicesto cheaper, energy-dense foods thatare easily obtained at conveniencestores, restaurants, and grocery stores.The current results suggest that suchlarger forces have a bigger role in theconsumption of unhealthy foodsthan in the consumption of healthyfoods.

The researchers also found thatdifferent aspects of the home foodenvironment are associated withhealthy and unhealthy food con-sumption in specific ways. Consistentwith the literature,32,33 the currentauthors found that availability andaccessibility of healthy foods andparental encouragement of healthyeating both reliably predicted higherintake of healthy foods. A healthyhome food environment predictedlower intake of unhealthy foods aswell, although the correlations weresmaller. Because the consumption ofunhealthy foods such as SSBs isassociated with the availability ofthose foods at home,34,35 it is possiblethat in the current sample theavailability and accessibility of fruitsand vegetables is negatively correlatedwith availability of SSB and snacks.However, because the measures todemonstrate this were not available inthe current dataset, this needs to beempirically verified in other data.

Although the measure of breakfastconsumption used here was limited(consumption at home, to be consis-tent with other home food environ-ment measures, as well as to avoidpotential confounding with SES-linked consumption of breakfast atschool), the data nevertheless re-vealed positive associations withhealthy eating and negative associa-tions with unhealthy eating. One un-expected finding in this study was therelative lack of association betweenfrequency of family meals and con-sumption of unhealthy foods. Theconstruct of family meals may beimprecise and mask substantial het-erogeneity in practice; for instance,they may occur at a table or aroundthe television. Recent research, forexample, showing that the effect offamily meals varies by race andethnicity, and has a detrimental effect

on obesity among Latino boys.36

Indeed, the construct of family itselfis nebulous, and family compositionlikely has large influences on what isidentified as a family meal.

This study has limitations. First,the data were from 2009 and maynot reflect current patterns of con-sumption. Second, nomeasures of un-healthy home food environment wereavailable for use in this study. Homeavailability and accessibility of itemssuch as SSBs, salty snacks, and dessertsare consistent predictors of consu-mption of these items.22,37,38 In theabsence of these measures, it cannotbe known with certainty whetherlower consumption of energy-dense,low-nutrient foods reflects low avail-ability of these foods in a contextwith high availability of vegetablesand fruit, or whether there is a truereduction in preference for thesefoods in such a context. Using mea-sures of unhealthy home environ-ments would have provided greaterclarity in understanding unhealthyeating among racial and ethnic mi-nority groups. Third, the data arecross-sectional, limiting inferencesregarding causality. It is possible thatparents structure the home environ-ment in response to children's obesityand food preferences. However, suchreverse causation would serve to biasresults toward the null; thus, these in-ferences regarding the influence of thefamily environment on diet are likelyconservative.

IMPLICATIONS FORRESEARCHAND PRACTICE

The large variations in the home foodenvironment found in this studyunderscore the necessity of takingrace and ethnicity into accountwhen examining the home food envi-ronment of children and adolescents.Both cultural and SES factors likelyhave a role in these differences inthe home food environment. Never-theless, the current results show thatimproving the home food envi-ronment (ie, increasing access andavailability of healthy foods) couldincrease the consumption of healthyfoods among racial and ethnic min-ority adolescents. In addition, it couldreduce the consumption of unhealthyfoods, although possibly not to

Journal of Nutrition Education and Behavior � Volume 47, Number 1, 2015 Ranjit et al 59

the same extent across racial andethnic groups. The consumption ofunhealthy foods among racial andethnic minorities may be responsiveto factors outside the home, such asthe larger food environment, as wellas to culturally specific factors.

Both cultural factors and unm-easured SES markers could accountfor some of these racial and ethnic dif-ferences. For example, economic inse-curity and limited time availabilityamong disadvantaged groups mayreduce the ability of the family to eatbreakfast together. Greater levels ofpsychosocial stress among econo-mically disadvantaged parents mayalso affect their ability to provideappropriate parenting behaviors suchas encouraging and reinforcinghealthy eating. Culturally mediatedlifestyles and practices also likelyhave a substantial role, possibly ininteraction with SES factors. Socialand cultural influences on food pre-ferences, shopping practices, time allo-cation, child-rearing strategies, andpriorities are all potentially importantdeterminants of home food availabil-ity, One study found that comparedwith African Americans, Hispanic fam-ilies purchased larger quantities offruits and vegetables; non-Hispanicwhites purchased more mixed dishesthanHispanics, and African Americanspurchased more protein foods thannon-Hispanic whites.12 In anotherstudy, acculturation and marital statuswere shown to be associated withlower availability and accessibility offruit and vegetables among His-panics.39

The findings reported in this studydo not address these larger culturalfactors, but they provide sufficientinsights into racial and ethnic differ-ences in diet and its responsiveness tothe home food environment, to use-fully informthedesignof interventionsto address such differences, as well asinform the conduct of further researchin the area. Further research in thisarea should specifically examine theprevalenceofunhealthyhomefooden-vironments amongminority racial andethnic groups, the extent to whichthese correlate negatively or positivelywith healthy home food environ-ments, and whether unhealthy dietsare more closely correlated with homefood environments or neighborhoodfood environments.

REFERENCES

1. Ogden CL, Carroll MD, Curtin LR,Lamb MM, Flegal KM. Prevalence ofhigh body mass index in US childrenand adolescents, 2007-2008. JAMA.2010;303:242-249.

2. Zhang Q, Wang Y. Using concentra-tion index to study changes in socio-economic inequality of overweightamong US adolescents between 1971and 2002. Int J Epidemiol. 2007;36:916-925.

3. Singh G, Siahpush M, Kogan M. Ris-ing social inequalities in US childhoodobesity, 2003-2007. Ann Epidemiol.2010;20:40-52.

4. Anderson SE, Whitaker RC. Preva-lence of obesity among US preschoolchildren in different racial and ethnicgroups. Arch Pediatr Adolesc Med. 2009;163:344-348.

5. Freedman DS, Khan LK, Serdula MK,Dietz WH, Srinivasan SR,Berenson GS. Racial differences in thetracking of childhood BMI to adult-hood. Obes Res. 2005;13:928-935.

6. Taveras EM, Gillman MW,Kleinman K, Rich-Edwards JW, Ri-fas-Shiman SL. Racial/ethnic differ-ences in early-life risk factors forchildhood obesity. Pediatrics. 2010;125:686-695.

7. Kirkpatrick SI, Dodd KW, Reedy J,Krebs-Smith SM. Income and race/ethnicity are associated with adherenceto food-based dietary guidance amongUS adults and children. J Acad NutrDiet. 2012;112:624-635.e626.

8. Arcan C, Kubik MY, Fulkerson JA,Story M. Sociodemographic differ-ences in selected eating practices amongalternative high school students. J AmDiet Assoc. 2009;109:823-829.

9. Chaloupka FJ, Powell LM. Price, avail-ability, and youth obesity: evidencefrom Bridging the Gap. Prev ChronicDis. 2009;6:A93.

10. Baker EA, Schootman M, Barnidge E,Kelly C. The role of race and povertyin access to foods that enable individualsto adhere to dietary guidelines. PrevChronic Dis. 2006;3:A76.

11. Kumanyika S, Grier S. Targeting inter-ventions for ethnic minority and low-income populations. Future Children.2006;16:187-207.

12. Cullen KW, Baranowski T, Owens E,Marsh T, Rittenberry L, de Moor C.Availability, accessibility, and prefer-ences for fruit, 100% fruit juice, andvegetables influence children‘s dietary

behavior. Health Educ Behav. 2003;30:615-626.

13. Kral TVE, Rauh EM. Eating behaviorsof children in the context of their familyenvironment. Physiol Behav. 2010;100:567-573.

14. Ventura A, Birch L. Does parentingaffect children’s eating and weight sta-tus? Int J Behav Nutr Phys Activity.2008;5:15.

15. Pearson N, Biddle SJH, Gorely T.Family correlates of breakfast con-sumption among children and adoles-cents: a systematic review. Appetite.2009;5:1-7.

16. MacFarlane A, Crawford D, Ball K,Savige G, Worsley A. Adolescenthome food environments and socioeco-nomic position. Asia Pacific J Clin Nutr.2007;16:748-756.

17. Affenito SG. Breakfast: a missed oppor-tunity. J Am Diet Assoc. 2007;107:565-569.

18. Wansink B, van Kleef E. Dinner ritualsthat correlate with child and adult BMI.Obesity (Silver Spring). 2013;22:E91-E95.

19. Birch LL. Development of food prefer-ences. Annu Rev Nutr. 1999;19:41-62.

20. Wardle J, Guthrie C, Sanderson S,Birch L, Plomin R. Food and activitypreferences in children of lean and obeseparents. Int J Obes Relat Metab Disord.2001;25:971-977.

21. Kuhl ES, Clifford LM, Stark LJ.Obesity in preschoolers: behavioralcorrelates and directions for treatment.Obesity (Silver Spring). 2012;20:3-29.

22. Neumark-Sztainer D. Preventingobesity and eating disorders in adoles-cents: what can health care providersdo? J Adolesc Health. 2009;44:206-213.

23. van der Horst K, Oenema A, Ferreira I,et al. A systematic review of environ-mental correlates of obesity-related die-tary behaviors in youth. Health EducRes. 2007;22:203-226.

24. Springer A, Kelder S, Byrd-Williams C, et al. Promoting energy-balance behaviors among ethnicallydiverse adolescents: overview and base-line findings of the Central TexasCATCHMiddle School Project.HealthEduc Behav. 2013;40:559-570.

25. Evans AE, Springer AE, Evans MH,Ranjit N, Hoelscher DM. A descriptivestudy of beverage consumption amongan ethnically diverse sample of publicschool students in Texas. J Am CollNutr. 2010;29:387-396.

26. Springer A, Lewis K, Kelder S,Fernandez M, Barroso C,

60 Ranjit et al Journal of Nutrition Education and Behavior � Volume 47, Number 1, 2015

Hoelscher D. Physical activity partici-pation by parental language use in 4th,8th, and 11th grade students in Texas,USA. J Immigr Minor Health. 2010;12:769-780.

27. Kuczmarski R, Ogden C, Guo S,et al. 2000 CDC growth charts forthe United States: methods and devel-opment. National Center for HealthStatistics. Vital Health Stat. 2002;11:1-190.

28. Hearn MD, Baranowski T,Baranowski J, et al. Environmental in-fluences on dietary behavior amongchildren: availability and accessibilityof fruits and vegetables enable con-sumption. J Health Educ. 1998;29:26-32.

29. Kratt P, Reynolds K, Shewchuk R.The role of availability as a moderatorof family fruit and vegetable consump-tion. Health Educ Behav. 2000;27:471-482.

30. Cullen KW, Baranowski T, Owens E,Marsh T, Rittenberry L, de Moor C.Availability, accessibility, and prefer-ences for fruit, 100% fruit juice, and

vegetables influence children’s dietarybehavior. Health Educ Behav. 2003;30:615-626.

31. Granner ML, Sargent RG,Calderon KS, Hussey JR, Evans AE,Watkins KW. Factors of fruit and vege-table intake by race, gender, and ageamong young adolescents. J Nutr EducBehav. 2004;36:173-180.

32. Jago R, Baranowski T, Baranowski JC.Fruit and vegetable availability: a microenvironmental mediating variable?Public Health Nutr. 2007;10:681-689.

33. Neumark-Sztainer D, Wall M,Perry C, Story M. Correlates of fruitand vegetable intake among adoles-cents: findings from Project EAT. Pre-vent Med. 2003;37:198-208.

34. EzendamNPM, Evans AE, StiglerMH,Brug J, Oenema A. Cognitive andhome environmental predictors ofchange in sugar-sweetened beverageconsumption among adolescents. Br JNutr. 2010;103:768-774.

35. Tak NI, Te Velde SJ, Oenema A, et al.The association between home envi-ronmental variables and soft drink

consumption among adolescents:exploration of mediation by individualcognitions and habit strength. Appetite.2011;56:503-510.

36. Rollins BY, Belue RZ, Francis LA.The beneficial effect of family mealson obesity differs by race, sex, andhousehold education: the NationalSurvey of Children’s Health, 2003–2004. J Am Diet Assoc. 2010;110:1335-1339.

37. Denney-Wilson E, Crawford D,Dobbins T, Hardy L, Okely AD. Influ-ences on consumption of soft drinks andfast foods in adolescents. Asia Pacific JClin Nutr. 2009;18:447-452.

38. Johnson L, van Jaarsveld CHM,Wardle J. Individual and family envi-ronment correlates differ for consump-tion of core and non-core foods inchildren. Br J Nutr. 2011;105:950-959.

39. Dave JM, Evans AE, Pfeiffer KA,Watkins KW, Saunders RP. Correlatesof availability and accessibility of fruitsand vegetables in homes of low-income Hispanic families. Health EducRes. 2010;25:97-108.