Appetite 53 (2009) 7683

Contents lists available at ScienceDirect

Appetite
journal homepage: www.elsevier.com/locate/appet

Research report

Food branding inuences ad libitum intake differently in children depending on weight status. Results of a pilot study
Jamie Forman a, Jason C.G. Halford b, Heather Summe a, Megan MacDougall a, Kathleen L. Keller a,*
a New York Obesity Research Center, St. Lukes-Roosevelt Hospital Center, Columbia University College of Physicians & Surgeons, 1090 Amsterdam Avenue 14A, New York, NY 10025, USA b School of Psychology, Eleanor Rathbone Building, Bedford Street South, University of Liverpool, Liverpool L69 7ZA, UK

A R T I C L E I N F O

A B S T R A C T

Article history: Received 9 March 2009 Received in revised form 11 May 2009 Accepted 15 May 2009 Keywords: Food marketing Food branding Childhood eating behavior Obesity Laboratory intake

Environmental changes have facilitated the rapid increase in childhood obesity. One such change is increased presence of food marketing which promotes intake of high-fat, energy-dense foods. This study tested the hypotheses that overweight (OW) children are more sensitive to the intake-enhancing effects of food branding than non-OW children, and that the relationship between weight status and intake of branded foods is mediated by level of food brand awareness. Forty-three non-OW (n = 23) and OW (n = 20) children from diverse ethnic backgrounds participated in four dinnertime visits to test their intake of meals where food brands were present (branded) or absent (unbranded). Food brand awareness was assessed by testing childrens abilities to match food brand logos with correct foods and name specic brands from recall. Weight and height were measured on the rst visit to determine BMI zscore and weight status. OW children consumed signicantly more energy per meal than non-OW. Child age and brand awareness were positively associated. OW children consumed an additional 40 kcal in branded vs. unbranded meals whereas non-OW children consumed 45 kcal less in branded meals. Overweight children showed greater responsiveness to food branding, and they may be at risk in environments that are highly inundated with messages about food. Published by Elsevier Ltd.

Introduction Currently, 17% of United States (US) children are considered obese and one in every three children is overweight (OW) (Ogden et al., 2006). While multiple biological factors may mediate individual susceptibility to obesity (Hill, 2006), most experts agree that environmental changes have facilitated the rapid increase of the disease in recent years. A primary environmental factor associated with childhood obesity is increased television (TV) viewing (Campbell, Crawford, & Ball, 2006; Cooper, Klesges, Debon, Klesges, & Shelton, 2006). Excess TV viewing increases the amount of sedentary time and might displace time available for physical activity. Additionally, TV viewing is often accompanied by food consumption (Matheson, Killen, Wang, Varady, & Robinson, 2004), particularly of energy dense snack foods (Coon, Goldberg, Rogers, & Tucker, 2001; Halford, Gillespie, Brown, Pontin, & Dovey, 2004) and this may be due in part to a disruption in food cue habituation that occurs when one is viewing a TV program (Temple, Chappel, Shalik, Volcy, & Epstein, 2007). Finally, food advertisements might directly affect intake by either stimulating hunger and/or encouraging

* Corresponding author. E-mail address: kk2092@columbia.edu (K.L. Keller). 0195-6663/$ see front matter . Published by Elsevier Ltd. doi:10.1016/j.appet.2009.05.015

children to consume the specic foods that are marketed (Borzekowski & Robinson, 2001). Most often foods advertised during childrens and family programs are cereals, snacks, and fast foods (Powell, Szczypka, Chaloupka, & Braunschweig, 2007), many of which are high in sugar, fat, and calories. Food marketing is considered one of multiple components of the external environment that show potential associations with obesity (Lobstein & Dibb, 2005), but the mechanism of this effect has not been investigated. Studies (Borzekowski & Robinson, 2001; Brody, Stoneman, Lane, & Sanders, 1981; Robinson, Borzekowski, Matheson, & Kraemer, 2007) suggest that food advertisements alter childrens preference for specic food brands, but few have investigated how advertisements subsequently impact food intake. One possibility is that advertisements act as cues for food consumption and exposure to them may promote eating in individuals who are responsive to these messages and/or images. Several studies suggest that OW adults (Herman & Polivy, 2008; Stoeckel et al., 2008) and OW children (Carnell & Wardle, 2008) may be more responsive to food cues and they often score higher on scales that measure external eating (Braet et al., 2008; Elfhag and Morey, 2008). If OW children are more motivated by external food cues, they might also be more vulnerable to inuences from food advertisements. Further, there is evidence that OW children nd food more reinforcing than non-OW children (Temple,

J. Forman et al. / Appetite 53 (2009) 7683

77

Bulkley, et al., 2008), so the combined forces of a memorable advertisement with a palatable food source could increase the temptation for children who are susceptible. Recent evidence to support this came from studies conducted by Halford et al. (2004, 2008). There were two main ndings in these studies. First, OW children were able to recognize more food-related advertisements than non-OW children (Halford et al., 2004). Second, OW children increased their food intake to a greater extent when they were exposed to food advertisements (Halford et al., 2004; Halford et al., 2008). Thus, under certain conditions, OW children appear more vulnerable to the effects of food advertising than non-OW children. However, the studies were conducted in older children (911 years) at an age where childrens experience and knowledge of food promotion was arguably already well developed. When these researchers examined children who were 57 years old, increases in intake after advert-exposure were similar regardless of weight status (Halford, Boyland, Hughes, Oliveira, & Dovey, 2007). Further studies to clarify these relationships in younger children are necessary. While researchers have predominately focused on the effect of a single advert or blocks of adverts on childrens intake, one question that remains unresolved is whether or not OW children are more vulnerable to other forms of food advertising, particularly branding. Branding is a form of marketing aimed at promoting recognition with a company brand or product, in the hopes that children will form emotional attachments to these products and eventually be lifelong consumers (Connor, 2006). Because children often reject foods that are novel or unfamiliar (Fallon, Rozin, & Pliner, 1984), product branding may help reduce this neophobia by offering children a familiar package for food delivery. The powerful effects of food branding were seen recently in a study by Robinson et al., where children had higher preferences for foods in McDonalds1 wrappers, regardless of whether the foods contained in the packaging came from McDonalds1 or not (Robinson et al., 2007). Because this study did not test for differences in brandings effects as a function of child weight status and did not assess the effects of branding on subsequent intake, additional studies are warranted. There were two main objectives for the present study. First, we tested the hypothesis that OW children ages 46 years will increase their intake under laboratory conditions where foods are branded compared to when foods are not branded, and this increase will be greater than the increase experienced by nonOW children. In order to determine the effect of child weight status on response to food brands independently, we adjusted for covariates such as age, sex, ethnicity, and socioeconomic status. Second, a novel instrument was developed using a logobased procedure to assess both brand recognition and recall to test the hypothesis that increased food brand awareness would be associated with higher BMI z-scores and would mediate the relationship between weight status and intake at the branded conditions. Methods Participants Forty-three, 46 years old children (26 girls; 17 boys), mean age (SD) 5.9 years (0.9) participated in this study. Children were from diverse ethnic backgrounds, with 42% African American, 19% Hispanic, 19% Caucasian, and 19% dened as other, which typically signied a child of mixed ethnic origin. Forty-ve percent of families were low-income, as identied by a reported annual earning of $20,000 per year, although 98% reported having cable television. Twenty-three children were classied as non-OW (BMI-for-age 85th %) and 20 were classied

as at risk for overweight (BMI-for-age ! 85th %), referred to as OW throughout. Four- to six years old children were selected because this age group has been shown to be highly affected by food advertisements and can recognize food brands (Fisher, Schwartz, Richards, Goldstein, & Rojas, 1991). The intention was to select children that were inuenced by food advertising but that were old enough to perform the testing procedures used to assess brand awareness. Families were recruited by posting yers on the Internet and in and around the study site (the Child Taste and Eating Laboratory at St. Lukes Hospital). Interested parents phoned the research staff to be screened, and if eligible, were scheduled for study visits. Children were excluded from the study if they had any pre-existing medical conditions, were taking any medications known to affect taste and body weight or did not have a television in the house. Also, eligible children had to be familiar with all the food brands used in the study, and had to report liking at least ve out of seven (assessed during the initial phone screening). Familiarity and liking for each of the seven foods were assessed during the initial phone screen by having the mother report the appropriate response for each food on a dichotomous scale (yes vs. no). Children who were not familiar with all seven of the food brands used in the test-meal were excluded. Also, children whose parents reported that they liked less than ve out of seven of the foods were also excluded. The purpose of this procedure was to ensure that unfamiliarity with the food brands would not affect childrens intake of these foods in the laboratory. The study was approved by the Institutional Review Board of St. Lukes Roosevelt Hospital. A parent (most often the mother) gave informed consent for the child, and the child verbally agreed to participate. After the study, each family was compensated $125 for time and travel expenses. Experimental design A 2 2 factorial design was used where non-OW and OW children (condition) were exposed to test meals where all foods were either branded or unbranded (treatment). Meals were presented to the children in randomized order across four separate, nonconsecutive visits to the Child Taste and Eating Laboratory. On two of the visits, children received meals in which the brands on all foods were clearly visible (branded), while on the other two visits, all foods were packaged in plain, unrecognizable plastic bags or containers (unbranded). Food brand recognition An age-appropriate instrument that used pictures of food logos to test childrens recognition and recall was developed that consisted of 30 pictures that were representative of food brand logos. Pictures and/or product logos were chosen that varied in familiarity and perceived healthfulness (see Appendix A). These foods were selected based on discussions between the lead investigators and research colleagues working on this project, and were designed to represent both wellrecognizable and lesser known food brand images in addition to brands from healthy and unhealthy foods. Because children this age may have a decreased ability to recall brand images by name (Macklin, 1983), a forced-choice procedure was established where each food brand logo was paired with 3 pictures of foods, one of which correctly matched with the logo. All pictures [a total of 30 brand logos, and 90 foods (3 per brand logo)] were laminated in 8 11.5 sheets and placed in a 3-ring binder for ease of presentation. One of the example presentations is found in Appendix B, for further explanation.

78

J. Forman et al. / Appetite 53 (2009) 7683

Food preparation and serving Prior to each visit, the following foods were prepared for each child: Kraft1 Lunchables Pizzas (Kraft Foods, Northeld, IL), Del Monte1 mixed fruit cups (Del Monte Foods, San Francisco, CA), Nabisco1 Oreo Cookies (Kraft Foods, Northeld, IL), Rold Gold1 pretzels (Frito Lay, Dallas, TX), Yoplait1 Trix Raspberry Rainbow/Strawberry Banana Bash avored yogurt (General Mills, Minneapolis, MN), Lays1 original potato chips (Frito Lay, Dallas, TX), Kool Aid1 cherry avored Jammers (Kraft Foods, Northeld, IL), and Nesquick1 chocolate milk (Nestle USA, Glendale, CA). For the branded conditions, all foods were pre-weighed in the packages and served to children on a tray, with the positions of each item kept constant across all visits. For the unbranded conditions, all foods were removed from the original packaging and placed in either clear plastic bags or small portion cups. For the pizzas, the cheese, pepperoni, and sauce were placed in 2-oz plastic portion cups, with weights for each recorded on the outside of the container. The pizza crust and candy from the Lunchables1 were wrapped individually in clear plastic bags. All drinks, including the one that came in the Lunchables1 package, were placed in clear, 8-oz plastic cups with lids and straws and were given to children at the time of serving. The chips, cookies, and pretzels were removed from original packages and served to children in plastic snack size sandwich bags. The fruit cups and yogurt were removed from the original package and placed in clear, plastic containers. Procedures Parents were instructed to have their children fast for at least 2 h prior to each visit. On the rst visit before dinner, mothers completed a brief questionnaire to assess the amount and types of TV shows children routinely watched. Also, brand awareness was assessed. Children were presented with pictures of each brand logo individually, followed by pictures of three possible food choices, and children were asked to choose the food that matches with the food brand. If children stated that they had never seen the brand, they were asked to make a guess. Once children selected a food picture, they were asked if they could name the food brand (e.g. McDonalds1, Coca Cola1, or Trix1). Thus, children had a possibility of receiving two points for each brand presented, one for choosing the correct food that matched with the brand and one for correctly naming the brand. With a total of 30 foods, scores on the brand awareness procedure ranged from 0 to 60. Testing facility and time Testing was completed at the Child Taste and Eating Laboratory located within St. Lukes Roosevelt Hospital, New York, NY. The laboratory contains two separate eating rooms where children ate dinner while a research assistant read books to the child to provide a non-food related distraction. All visits took place at dinner time, between 4 and 6:30 pm, and were completed within 1 h, with 30 min allotted for dinner. Anthropometrics On the rst visit, children were weighed and measured in light clothing and without shoes on a balance scale and stadiometer. Weights were recorded to the nearest 0.5 lb and height to the nearest 0.25 in. All weights and heights were converted to body mass indices (BMIs = kg/m2) and BMI z-scores using the CDC conversion program for SAS version 9.0.

Test-meal intake Children were allowed 30 min to eat ad libitum from their respective dinners, and they were instructed to let a researcher know if they desired additional servings. Because children this age are often too shy to make these requests, the researchers routinely asked children if they wanted seconds after they nished an item. During dinner, children were read non-food related stories by research staff to provide neutral distractions. While a few of the children mentioned that the foods looked the same as last time, none of them guessed the purpose of the study. If children asked questions that pertained to the nature of the experiment, the research staff was instructed to divert their attention away from the brand-related inuences of the food by refocusing the childs attention on the story that was being told. Power analysis A power analysis was done prior to the study based on data from Halford et al. (2004). Using a two-tailed signicance test and a p-value of 0.05, it was determined that with 44 subjects, we would have 90% power to detect a mean difference between groups as small as 28 kcal. Statistical analyses Descriptive statistics (means and SDs) were computed on all main study variables that were continuous (e.g. child age, BMI z-scores, brand recognition scores, and meal intakes). Frequencies were computed for all categorical variables (e.g. child sex and ethnicity). Total intake from the four meals was computed from the gram quantities of each food consumed by using information from product food labels. For meal-related continuous variables, means and SDs were computed across the two branded and two unbranded conditions, and across all four conditions together. In addition to total meal energy (kcals), the amount (means and SDs) of energy consumed from the individual foods at each meal was computed. The primary study objective was analyzed in two ways. First, a difference score was created for intake at the branded unbranded conditions, such that higher scores depict greater intake when foods are branded. Independent samples T-tests were used to determine if OW children had greater difference scores than non-OW children, or in other words, if OW children show an increased consumption when foods are branded, compared to non-OW children. Two-way ANOVA were used to test for differences in non-OW and OW children across the treatment conditions (branded vs. unbranded). MANOVA were done to test the effects of sex, ethnicity, SES, and other variables of interest on the above analyses. Where appropriate, Tukeys post hoc tests were used to identify between-group differences. In regards to the secondary objectives, Pearsons correlation coefcients were calculated to assess the relationships between brand awareness, child BMI z-score, child age, and intake at the test meals. In addition, to determine if brand awareness mediates the relationship between child weight status and intake, partial correlation coefcients were computed, with brand awareness as a covariate, between child BMI z-score and intake at the branded conditions. For all statistical tests, a p-value of less than or equal to 0.05 was the cut-off for signicance. Descriptive statistics were reported as means S.D. All analyses were done using SPSS version 16.0 for Windows XP.

J. Forman et al. / Appetite 53 (2009) 7683 Table 1 Descriptive statistics. Variable BMI-for-age < 85th % (n = 23) % (n) Gender %Boys %Girls Ethnicity African-American Hispanic Caucasian Asian Other Yearly Income <20,000 !20,000 Age BMI z-scorec Total intakec
a b c

79

Table 2 Description of television viewing related behaviors. BMI-for-age ! 85th % (n = 20) % (n) Reported having cable Yes No Reported total hours TV/Week 15 or less >15 Brand awareness
a

Variable

BMI-for-age < 85th % (n = 23) % (n)

BMI-for-age ! 85th % (n = 20) % (n)

30.4 (7) 69.6 (16)

50.0 (10) 50.0 (10)

91.3 (21) 8.7 (2)

90.0 (18) 10.0 (2)

39.1 21.7 17.4 8.7 13.0

(9) (5) (4) (2) (3)

35.0 40.0 15.0 0.0 10.0

(7) (8) (3) (0) (2)

60.9 (14) 39.1 (9) 32.7 6.8a

65 (13) 35 (7) 34.6 6.6a

Mean SD.

30.4 (7) 69.6 (16) 5.9 0.9b 0.01 1.1b 368.8 139.3b

45.0 (9)a 50.0 (10) 5.9 0.9b 1.5 0.7b 559.7 185.9b

Missing data due to one family who did not respond to this question. Mean SD. Signicantly different at p 0.05.

Statement of ethics We certify that all applicable institutional and governmental regulations concerning the ethical use of human volunteers were followed during this research. Results In total, 48 children were enrolled in the study. Due to scheduling conicts, four families were not able to make the rst visit and were dropped from the study. One family decided to withdraw after the rst visit because of difculty scheduling the follow-up visits. A total of 43 children completed all 4 visits, and data for those children were used for the remaining analyses. There were no differences among the main outcome variables between those that completed the study and those that did not. Descriptive statistics for children are displayed in Table 1. There were no differences between non-OW and OW children for age (p = 0.9), sex (p = 0.3) or ethnic group breakdown (p = 0.5). Across all meals (regardless of condition), OW children consumed nearly 200 kcal more than non-OW children (t = 3.8; p < 0.001). There were no differences as a function of weight status for reported number of hours the children spent watching TV per week (p = 0.3). Overall, most parents reported that their children watched between 8 and 15 h of TV/week (48.8%) or 16 and 23 h of TV/week (30.2%). There were no differences in reported hours spent watching TV between gender or ethnic groups. Brand awareness score did not differ as a function of child weight status (p = 0.4) (Table 2). Brand awareness and recognition Total brand awareness score (out of 60) was 33.6 6.7. Children had higher awareness of familiar food brands than unfamiliar brands (65.5 14.2% vs. 46.4 9.7%, for familiar and unfamiliar brands, respectively). Brand awareness of unhealthy foods was higher than awareness for healthy/neutral foods (60.0 10.0% vs. 52.9 13.2, respectively). There were no differences in recognition of healthy foods (p = 0.1) or unhealthy foods (p = 0.9) as a function of child weight status. For the foods used in the brand awareness instrument, parents were asked whether they regularly purchase these products at

home so we could evaluate if the childs experience with these products affected their ability to recognize them in the laboratory. Parental report of using these foods in the home was inversely, though not signicantly, associated with the childs total brand awareness score (r = 0.2; p = 0.1). For familiar foods, parental report of using the product at home was negatively associated with childs awareness for that product in the laboratory (r = 0.3; p 0.05). For unfamiliar foods, parental report was not correlated with childs awareness for the product in the laboratory (p = 0.7). Child age was positively associated with brand awareness score (r = 0.6; p 0.001). Brand awareness was not associated with intake at the branded (p = 0.3) or the unbranded (p = 0.7) conditions and adjusting for age did not change these associations. Total brand awareness was also not associated with BMI z-score (p = 0.7). Ad libitum food intake in response to food brand presentation Intake as a function of meal condition (branded vs. unbranded) and child weight status (non-OW vs. overweight) is depicted in Fig. 1. Overall, there were no differences in intake at the branded vs. unbranded conditions (p = 0.8), and the interaction between meal condition and weight status was also not signicant (p = 0.2). Adjusting for age, sex, and brand awareness did not change the above relationships (Fig. 1).

Fig. 1. Intake of meals in the branded (striped box) and unbranded (black box) conditions in all children, and broken down by non-OW and OW. There were no differences as a function of treatment group in all children (p = 0.8), or as a function of weight status (p = 0.2).

80

J. Forman et al. / Appetite 53 (2009) 7683

The order in which children received the dinners (branded rst vs. unbranded rst) did not signicantly affect intake across all children (p = 0.7). However, order and weight status (non-OW vs. OW) interacted to differentially inuence intake in children. Overweight children who received the branded condition rst consumed an average of 567.8 29.5 kcal across all four meals. This was over 250 kcal more than intake for non-OW children who were exposed to the branded condition rst (mean intake = 301.8 32.3) (F(1,41) = 11.0; p = 0.001). Tukeys post hoc tests revealed that OW children who received the branded meal rst also consumed more than both non-OW and OW children who were exposed to the unbranded meal rst (420.3 32.3 and 477.0 32.3 kcal, respectively). Food items that were served at the meal were analyzed individually to determine if meal condition (branded vs. unbranded) differentially inuenced intake. There was a nonsignicant trend for all children, regardless of weight status, to consume more Nesquick1 Chocolate milk at the unbranded condition (t = 1.8; p = 0.07). For all other foods, there were no differences by meal condition. The difference score between the branded and unbranded conditions is displayed in Fig. 2. Non-OW children ate an average of 45.3 kcal less at the branded condition (difference score = 45.3 23.4), while OW children ate an average of 40.7 kcal more at the branded condition (difference score = 40.7 30.2) [t = 2.3; p < 0.05]. After adjusting the dependent variable (difference score) for brand awareness, the p-value was slightly decreased (p = 0.04). Adjusting the difference score simultaneously for ethnicity and income did not change the overall relationship, but after adjusting for sex, the p-value was no longer signicant (p = 0.06). Boys tended to increase their intake at the branded condition to a greater extent than girls did (t = 2.1, p < 0.05) (Fig. 2). Discussion The objective of this study was to investigate the inuence of popular food brands on laboratory intake in 46 years old children. Overall, OW children ate more across both conditions consistent with expectations validating this experimental approach to studying factors that may provoke hyperphagia. While not all original hypotheses were supported, the results suggest that OW children may react differently to food branding than non-OW, and follow up studies may want to investigate the sources of these differences. Particularly, children above the 85th BMI-for-age percentile consumed about 40 kcal more in the presence of popular food brands; non-OW children, on the other hand, actually consumed over 40 kcal less when foods were branded. These differences were in part explained by differences in food brand

Fig. 2. Independent T-test results comparing the difference score (DS) (intake at the branded conditionintake at the unbranded condition) between non-OW and OW children. OW children ate 40.7 kcal more at the branded condition (DS SD = 40.7 30.2 kcal) while non-OW children ate 45.4 kcal less (DS SD = 45.3 23.4 kcal) [t = 2.3; p < 0.05]. *Signicantly different at p 0.05.

responsiveness as a function of child sex; boys ate about 45 kcal more at the branded conditions, while girls ate about 40 kcal less. This divergence of response to branding in the differing weight and sex groups suggests that susceptibility to external cues from food packaging may not be equal among children. Such hyperresponsiveness may be critical to hyperphagia and the development of obesity in children, and future studies to conrm these ndings in larger populations are warranted. The relationship between intake and meal condition observed in the present study did not appear to be mediated by the childrens brand awareness, although there were limitations to the instrument that will be discussed below. Nonetheless, determining whether overweight children respond more to brand images because of greater previous exposure to advertising and marketing is an important etiological question. In the present study, the parents of the overweight children did not report greater TV viewing. However, given that larger scale studies demonstrate a link between TV viewing time and weight status (Anderson, Crespo, Bartlett, Cheskin, & Pratt, 1998; Crespo et al., 2001), this possibility should not be eliminated. Moreover, the association between parental reports of purchasing the food brands and childrens awareness of these brands in the laboratory was negative, particularly for brands that were more familiar (e.g. McDonalds1, Trix1 and M&Ms1). This suggests that the home feeding environment is not the primary site where children learn their recognition and awareness of food brands. Future interventions aimed at reducing food advertising exposure among children should take this into account. Our initial prediction was that all children, regardless of weight status, would consume more during branded meals, but based on Halford et al. (2004, 2008) we hypothesized that OW children would show greater food brand responsiveness. This prediction was based on the knowledge that children are initially apprehensive to try foods that are new or unfamiliar (Fallon et al., 1984; Pliner & Hobden, 1992), and the presence of familiar brands would theoretically remove that apprehension. Surprisingly, while the OW children did increase their intake in the branded condition, the non-OW actually showed reduced intake when foods were branded. We intentionally selected brands that were targeted at children (e.g. Lunchables1 and Trix1 yogurt) and ensured that those enrolled were familiar with these products. However, it is possible that children had different levels of previous experience with the brands used in the test meals and this may have had an impact on their intake of these foods. While neither brand awareness nor reported television media exposure appeared to differ, the fact that non-OW children showed less consumption at the branded meals may indicate familiarity with the actual brand, but a lack of familiarity with actually consuming it. It is possible the non-OW children have previously not been presented with these specic foods in their packaging, and thus, may have been less familiar with consuming them in that form. In addition, these results may also be reective of differences in family eating habits between OW and non-OW children that we were not able to assess in the current design. Alternatively, accessibility of the food options in the laboratory may have also affected intake at the meal conditions. The foods at the unbranded conditions were served in small plastic containers and re-sealable clear plastic bags. Some children may have felt more comfortable sampling from foods served in these more accessible containers, particularly if they only wanted to taste a small portion of the food. For one food in particular, the Nesquick1 chocolate milk, there was a trend for all children, regardless of weight status, to consume more in the unbranded condition. Serving the drink in a clear 8-oz plastic cup with a straw may have made consumption easier for the children, regardless of weight status, compared to serving it in the 20-oz container in which it is sold. While we offered children a straw in both conditions, the

J. Forman et al. / Appetite 53 (2009) 7683

81

larger container may have been harder for them to handle. Future studies are warranted to clarify these ndings. In hindsight it may have been better to provide small, size-matched portions of branded and unbranded foods, but with the exception of some items, branded foods are not always sold in single portion/serving units making such experiments difcult to control. To rule out the possibility that order of exposure to the treatment (i.e. branded rst vs. unbranded rst) may have inuenced intake at subsequent meals, we tested for order effects in the ANOVA. There were no overall order effects, but it is interesting to note that weight status and presentation order showed a signicant interaction. Overweight (OW) children who were presented with the branded meal rst averaged more intake across meals than did the other groups (i.e. OW who received the unbranded rst, and non-OW who received either branded or unbranded rst). This did not appear to be due to overeating in OW children at the rst branded meal. Given that this was not the focus of this study, additional explanations about this nding would be, at best, speculative. We chose to use a combination of two methods to test brand awareness; matching (or recognition) and recall. Brand recognition has been used before (Fisher et al., 1991) with similar aged children, and results suggest it is appropriate for the age group studied. However, it is possible that the recall exercise was too difcult for some children in the study. When the recognition and recall scores were computed separately, the mean recognition score was 26.4 out of 30, while the mean score for recall was only 7.2 out of 30. In addition, brand recall was more strongly associated with child age (p < 0.001), than recognition was (p < 0.05), suggesting that child age was an important factor in determining childrens ability to perform in the recall task. These observations are supported by Macklin (1983) who found that preschool aged children had a more difcult time recalling brand names than they had recognizing them. Recall involves greater cognitive processing than recognition, and preschool aged children tend to have a difcult time searching for and recalling specic memories (Macklin, 1983). Furthermore, by the age of kindergarten, children are able to recall brand names of common food products, but this ability is not extensive (Ward, Wackman, & Wartella, 1977). Future studies using this method might choose to eliminate the recall exercise in favor of a matching exercise using a wider variety of foods. An additional factor that may have inuenced childrens food intake, but was not assessed, was the reinforcing value of the foods themselves. Recent studies suggest that daily consumption of a palatable snack food across a 2-week period decreases the extent to which adults nd the food reinforcing (Temple, Giacomelli, Kent, Roemmich, & Epstein, 2007). However, follow-up studies in obese women demonstrated that daily snack consumption led to an increase in the reinforcing value of the food, not a decrease as seen in non-obese controls (Temple, Legierski, Giacomelli, Salvy, & Epstein, 2008). Because children consumed the same foods across four visits, though they were non-consecutive, it is possible that the desirability of the foods changed across the study, and that this impacted the results. Because the reinforcing value of foods can change with level of exposure, future studies might wish to adjust for previous taste exposure to account for this. There were several strengths to this study that should be discussed. First, even though this was a pilot investigation, the results suggest that some children may be more responsive to food brand images than others. Future studies should be planned to better identify the factors that contribute to increased vulnerability to food brand images in this population. Once these issues are claried, these ndings can be used to promote public policy changes in the marketing of foods to young children. The diversity of the children in the study can be viewed as both a strength and a

limitation. There is a paucity of studies that have reported direct measures of food intake in ethnically diverse groups of children. However, over 1/3 of the study population was Hispanic, and though all these children spoke English as their rst language, their parents/families often spoke Spanish to them in the home. This may have inuenced their performance on the brand awareness instrument, particularly because it is not known how many of our families used Spanish words to refer to these food products in the home. The use of a controlled feeding laboratory to measure food intake can also be viewed as a strength. While laboratory intake may not reect long-term or usual intake, it offers a highly controlled atmosphere where food packaging cues can be manipulated and outcomes measured. Future studies are warranted to conrm these ndings in a more naturalistic eating environment. In addition to the previously discussed issues with the brand awareness instrument, there were a few other limitations that should be noted. First, the sample size was small, and because this was a pilot study, results should be interpreted with caution. We were powered to detect differences between two groups, but we did not have sufcient power to test interactions between weight status and branding condition. Further, the choice of foods to use in both the brand awareness instrument and the meal was somewhat arbitrary and may have affected the results. Our intention was to select foods that were targeted at children and were easy to eat without additional preparation. In doing so, it is probable that some enticing food brands that are highly appealing to children in the study were not used. Thus, it is not possible to generalize these ndings to other foods and/or meal conditions. Finally, the use of a multi-item meal may have been distracting because each of the food brands can affect children differently. For example, the Trix1 yogurt advertised a contest with monetary prizes where every 1 in 50,000 yogurts contain a winning sticker inside. Because most children could read this, the effect of the branded yogurt on intake might expectedly have been greater than foods that did not offer such premiums. Future studies should control these differences across food brands to better understand the components of food packaging that affect intake. In conclusion, these pilot data reveal that weight status inuences childrens responsiveness to food branding. Even though overweight children did not have higher brand awareness scores, they did increase their intake by about 40 calories during meals where food brands were exposed. While this increase is small, if it is sustained daily across 3 meals, it is possible to amass an additional 12 pounds over the course of a year. However, the implications to these ndings should be interpreted with caution because the sample size was small and the brands tested were diverse. Nonetheless, some children may be more vulnerable to the effects of food branding than others. Developing measures to screen for this vulnerability would be a worthwhile research endeavor. If subsequent research conrms the link between branding, food choice and caloric intake, these ndings may have major implications for the marketing of unhealthy foods to children. Acknowledgements This research was supported by NIH grant K01DK068008 (KLK) and by a Pilot and Feasibility Grant awarded from the New York Obesity Research Center. Also, the work was made possible by the Obesity Research Center Grant (NIH grant 5P30DK026687-27). Research was conducted primarily by J. Forman, who was at the time an Intel Science Research student enrolled in high school. She was nationally recognized for this work. K. Keller, the PI on the study, contributed to study design, data collection, analysis, and writing.

82

J. Forman et al. / Appetite 53 (2009) 7683 Halford, J. C., Boyland, E. J., Hughes, G., Oliveira, L. P., & Dovey, T. M. (2007). Beyondbrand effect of television (TV) food advertisements/commercials on caloric intake and food choice of 57-year-old children. Appetite, 49, 263267. Halford, J. C., Boyland, E. J., Hughes, G. M., Stacey, L., McKean, S., & Dovey, T. M. (2008). Beyond-brand effect of television food advertisements on food choice in children: the effects of weight status. Public Health Nutrition, 11, 897904. Herman, C. P., & Polivy, J. (2008). External cues in the control of food intake in humans: the sensory-normative distinction [Review] [49 refs]. Physiology & Behavior, 94, 722728. Hill, J. O. (2006). Understanding and addressing the epidemic of obesity: an energy balance perspective. Endocrine Reviews, 27, 750761. Lobstein, L., & Dibb, S. (2005). Evidence of a possible link between obesogenic food advertising and child overweight. Obesity Reviews, 6, 203208. Macklin, C. M. (1983). Do children understand TV ads? Journal of Advertising Research, 23(1), 6370. Matheson, D. M., Killen, J. D., Wang, Y., Varady, A., & Robinson, T. N. (2004). Childrens food consumption during television viewing. American Journal of Clinical Nutrition, 79, 10881094. Ogden, C. L., Carrol, M. D., Curtin, L. R., McDowell, M. A., Tabak, C. J., & Flegal, K. M. (2006). Prevalence of overweight and obesity in the United States 19992004. Journal of the American Medical Association, 295, 15491555. Pliner, P., & Hobden, K. (1992). Development of a scale to measure the trait of food neophobia in humans. Appetite, 19, 105120. Powell, L. M., Szczypka, G., Chaloupka, F. J., & Braunschweig, C. L. (2007). Nutritional content of television food advertisement seen by children and adolescents in the United States. Pediatrics, 120(3), 576583. Robinson, T. N., Borzekowski, D. L., Matheson, D. M., & Kraemer, H. C. (2007). Effects of fast food branding on young childrens taste preferences. Archives of Pediatrics & Adolescent Medicine, 161, 792797. Stoeckel, L. E., Weller, R. E., Cook, E. W., III, Twieg, D. B., Knowlton, R. C., & Cox, J. E. (2008). Widespread reward-system activation in obese women in response to pictures of high-calorie foods. Neuroimage, 41, 636647. Temple, J. L., Chappel, A., Shalik, J., Volcy, S., & Epstein, L. H. (2007a). Daily consumption of individual snack foods decreases their reinforcing value. Eating Behaviors, 9, 267276. Temple, J. L., Giacomelli, A. M., Kent, K. M., Roemmich, J. N., & Epstein, L. H. (2007b). Television watching increases motivated responding for food and energy intake in children. American Journal of Clinical Nutrition, 85, 355361. Temple, J. L., Bulkley, A., Badawy, R., Krause, N., McCann, S., & Epstein, L. H. (2008a). Differential effects of daily snack food consumption on food reinforcement in obese and non-obese women. Obesity, 16, S67 (Abstract). Temple, J. L., Legierski, C. M., Giacomelli, A. M., Salvy, S. J., & Epstein, L. H. (2008b). Overweight children nd food more reinforcing and consume more energy than do nonoverweight children. American Journal of Clinical Nutrition, 87, 11211127. Ward, S., Wackman, D. B., & Wartella, E. (1977). How children learn to buy: the development of consumer information-processing skills. Beverly Hills, CA: Sage.

References
Anderson, R., Crespo, C., Bartlett, S., Cheskin, L., & Pratt, M. (1998). Relationship of physical activity and television watching with body weight and level of fatness among children. Journal of the American Medical Association, 279, 938942. Borzekowski, D. L. G., & Robinson, R. N. (2001). The 30-second effect: an experiment revealing the impact of television commercials on food preferences of preschoolers. Journal of the American Dietetic Association, 101(1), 4246. Braet, C., Claus, L., Goossens, L., Moens, E., Van, V. L., & Soetens, B. (2008). Differences in eating style between overweight and normal-weight youngsters. Journal of Health Psychology, 13, 733743. Brody, G. H., Stoneman, Z., Lane, T. S., & Sanders, A. K. (1981). Television food commercials aimed at children, family grocery shopping, and motherchild interactions. Family Relations, 30, 435439. Campbell, K. J., Crawford, D. A., & Ball, K. (2006). Family food environment and dietary behaviors likely to promote fatness in 56 year-old children. International Journal of Obesity, 30, 12721280. Carnell, S., & Wardle, J. (2008). Appetite and adiposity in children: evidence for a behavioral susceptibility theory of obesity. American Journal of Clinical Nutrition, 88, 2229. Connor, S. M. (2006). Food-related advertising on preschool television: building brand recognition in young viewers. Pediatrics, 118, 14781485. Coon, K. A., Goldberg, J., Rogers, B. L., & Tucker, K. L. (2001). Relationships between use of television during meals and childrens food consumption patterns. Pediatrics, 107, E7. Cooper, T. V., Klesges, L. M., Debon, M., Klesges, R. C., & Shelton, M. L. (2006). An assessment of obese and non obese girls metabolic rate during television viewing, reading, and resting. Eating Behaviors, 7, 105114. Crespo, C. J., Smit, E., Troiano, R., Bartlett, S., Macera, C., & Anderson, R. (2001). Television watching, energy intake, and obesity in US children: results from the Third National Health and Nutrition Examination Survey 19881994. Archives of Pediatrics & Adolescent Medicine, 3, 260265. Elfhag, K., & Morey, L. C. (2008). Personality traits and eating behavior in the obese: poor self-control in emotional and external eating but personality assets in restrained eating. Eating Behaviors, 9, 285293. Fallon, A. E., Rozin, P., & Pliner, P. (1984). The childs conception of food: the development of food rejections with special reference to disgust and contamination sensitivity. Child Development, 55, 566575. Fisher, P. M., Schwartz, M. P., Richards, J. W., Goldstein, A. O., & Rojas, T. H. (1991). Brand logo recognition by children aged 3 to 6 years: Mickey Mouse and Old Joe the Camel. Journal of the American Medical Association, 266, 3145 3148. Halford, J. C., Gillespie, J., Brown, V., Pontin, E. E., & Dovey, T. M. (2004). Effect of television advertisements for foods on food consumption in children. Appetite, 42, 221225.

Appendix A. Food brand logos used in brand recognition instrument

Brand Logo Trix Rabbit Aunt Jemima Hersheys Kiss Planters Peanut Teddy Graham (TG) Green Giant M&Ms Snap, Krackle, & Pop (Rice Krispies) Hamburger Helper Spaghettios Tony the Tiger Chef Boyardee Kool-Aid Man McDonalds Cocoa Puffs Pringles Lucky Charms Keebler Elf Quaker Oats KFC Count Chocula Hawaiian Punch Capn Crunch Coca-Cola Pillsbury Starkist Tuna Goldsh Chester Cheeta Sunmaid Raisins Oscar Meyer Weiner Food choices presented Cereal, apple, wafes Cereal, apple, pancake/syrup Ice cream, candy, chocolate Chips, peanuts, pretzels TG cookies, crackers, chocolate chip cookies Fruit, vegetables, chicken M&M candy, sour candy, jelly beans Banana, raspberries, cereal Lasagna, spaghetti, chicken Noodles, green beans, pasta Milk, cereal, wafes Chicken, steak, spaghetti OJ, fruit drink, soda Hamburger, hotdog, taco Pancakes, strawberry, cereal Cookies, chips, pretzels Cereal, apples, orange Chips, cookies, pretzels Raspberry, wafe, oatmeal Tacos, fried chicken, hamburger Berries, cereal, granola bar Milk, OJ, fruit punch Wafes, orange, cereal Soda, juice, milk Pasta, cinnamon roll, pizza Tuna, soup, bagel with lox Pretzels, chips, sh crackers Peanuts, crackers, cheese curls Raisins, oranges, apples Taco, hamburger, hotdog Perceived familiarity High Low High Low High Low High High Low Low High Low High High Low Low High High Low High Low Low High High Low Low High High Low Low Perceived healthfulness Neutral Low Low High Neutral High Low High Neutral Neutral Neutral Neutral Low Low Neutral Low Neutral Low High Low Neutral Low Neutral Low Low High Neutral Low High Low

J. Forman et al. / Appetite 53 (2009) 7683

83

Appendix B. Sample question from brand recognition instrument Instructions: Show children the picture of the food brand logo followed by the three pictures of foods. Ask questions 1 and 2 below.

Question 1: Does this logo (point to the logo) match with the hamburger and French fries (point to hamburger and fries), the hotdog (point to hotdog), or the tacos (point to tacos). Question 2: Can you tell me the name of this brand? (answer: McDonalds).