|Home | About | Journals | Submit | Contact Us | Français|
The effect of television viewing (TVV) with and without advertisements (ads) on energy intake is unclear.
To test: 1) the effect of TVV, with and without ads, on energy intake compared to a control and reading condition, and 2) the association of distractibility and memory for ads with energy intake and body weight.
Forty-eight (26 female) adults (age 19 to 54 years) with body mass index 20 to 35 kg/m2 completed this laboratory-based study. All participants completed four buffet-style meals in random order: 1) control, 2) reading, 3) TV-with food and non-food ads (TV-ads), 4) TV-no ads. Energy intake was quantified by weighing foods. Distractibility and memory for ads in the TV-ads condition were quantified with a norm-referenced test and recognition task, respectively.
Repeated measures analysis of variance indicated that energy and macronutrient intake did not differ significantly among the four conditions (p-values>0.65). Controlling for sex, memory for advertisements was associated with body weight (r=0.36, p<0.05) and energy intake, but only when viewing TV (r=0.39, p<0.05 during the TV-no ads condition and r=0.29, p=0.06 during the TV-ads condition). Controlling for sex, distractibility was associated with body weight (r=0.36, p<0.05), but not energy intake. Distractibility, however, accounted for 13% of the variance in men's energy intake (p=0.11).
TVV did not affect energy intake, but individual characteristics (memory for advertisements) were associated with body weight and energy intake in certain conditions. These characteristics should be considered in food intake and intervention studies.
Longitudinal studies indicate that television viewing (TVV) is associated with increased body mass in adults (1) and adolescents (2), suggesting that TVV affects energy intake, energy expenditure (activity levels), or both. Most laboratory-based studies indicate that TVV increases energy intake during test meals (3-6), though participant characteristics, study sample size, and effect sizes vary considerably [TVV has been found to increase energy intake by 11.5% (3) to 71% (5)]. At least one study, however, found that TVV decreased the energy intake of children (7).
Television viewing might increase energy intake through a number of mechanisms, including the effect of advertisements (ads) on behavior, mood alteration, or by fostering passive over-consumption and dishabituation to food cues. Exposure to food ads increased the energy intake of children and adolescents (8, 9), and induction of positive and negative emotional arousal with films increased women's energy intake with increasing levels of dietary restraint (10, 11). Television viewing might also increase food intake by decreasing awareness of the amount of food that is being ingested (12), or by disrupting habituation to food cues (13). Temple et al. (13) found that TVV disrupted habituation to food cues only if TVV required attention allocation. Consequently, people who score high on measures of distractibility and whose attention is easily captured by external stimuli should experience larger increases in energy intake during TVV. In addition, the presence of other distracting stimuli, such as reading, could also increase energy intake among individuals with high levels of distractibility. Importantly, passive distracting stimuli such as TVV contribute to energy intake more so than distracting tasks that require visual-motor resources, such as playing video games (14).
This is one of the first studies with adult participants to directly examine: 1) the effect of TVV, with and without ads, on energy intake; 2) if distractibility is associated with energy intake or body weight; and 3) the influence of memory for ads on energy intake or body weight. It was hypothesized that energy intake would be greater when viewing: 1) TV with and without ads compared to a control and reading condition, and 2) TV with ads compared to viewing TV without ads. It was also hypothesized that distractibility would be associated with energy intake and body weight and, based on previous studies conducted with children (9, 15), that memory for ads would be associated with energy intake.
The study was conducted at the Pennington Biomedical Research Center (PBRC), Baton Rouge, LA, USA. All applicable institutional and government regulations concerning the ethical use of human volunteers were followed. All participants provided written informed consent and the research was approved by the Institutional Review Board of the PBRC.
Fifty-one healthy males and females, ages 18 to 54 years with body mass index 20 to 35 kg/m2, inclusive, were enrolled in the study. Exclusion criteria were: 1) use of medications that affect eating behavior or body weight (e.g., antipsychotic medication); 2) diagnosis of a chronic disease such as diabetes, cardiovascular disease or cancer; 3) tobacco use; 4) refusal to eat the foods provided during the study; and, for females, 5) irregular menstrual cycles or pregnancy. Three participants failed to complete the study due to scheduling difficulties. Participants received monetary compensation ($125 USD) for participation.
Using a within subjects or repeated measures design, participants completed four conditions in random order. During all conditions, participants were provided with a meal consisting of 16 food items served simultaneously in individual dishes that included grilled and breaded chicken and other low- and high-fat food items. Water and fruit punch were also served with each meal. The types and amount of foods served during the test meals are outlined in Table 1. The four conditions follow.
Participants completed two days of testing, consuming a standard 359 kcal breakfast followed by the first test meal at lunch (4 hours after breakfast) and the second test meal at dinner (4.5 hours after lunch). The food items and energy content of the breakfast, lunch, and dinner meals are provided in Table 1. On the second test day, the same standardized breakfast was consumed, followed by the third test meal at lunch and the fourth test meal at dinner. Test days were scheduled 7 ± 2 days apart. To reduce demand characteristics, participants were told that the purpose of the study was to test the consistency of taste rating over time. Energy intake was quantified by weighing each food item before serving it to participants and after participants ate. Each food was measured to 0.1 grams and the food was weighed out of sight of the participant.
The study relied on a within-subjects design; as such, each participant served as his/her own control and all participants completed all four conditions. The sample size was determined by an a priori power analysis that relied on the published literature to estimate the minimally acceptable effect size (10.75% increase in energy intake). Variance estimates for energy intake were calculated based on previously collected laboratory data that were derived from similar test meals. The power analysis assumed a large standard deviation (SD) for energy intake (465 kcal), 80% power, an effect size of 10.75%, alpha = 0.05, and a one-sided hypothesis test. Given these assumptions, the results of the power analysis indicated that we could detect a 10.75% increase in energy intake with 48 participants. This power estimate was considered conservative based on the large variance and inclusion of an effect size that is smaller than any effect size reported in the literature to date.
During the TV conditions, participants viewed two different episodes of the same program (Net Cafe, http://www.archive.org/details/Performi1999). This program describes the uses of the internet for security technology, arts, and entertainment, including the performing arts. The Net Café program was broadcast in over 100 countries. The production style and episode templates were consistent between the two episodes, with both focusing on interviews and demonstrations with four to five technology developers and community members. One episode contained ads and one did not. The ads consisted of six food ads (e.g., a pizza commercial) and six non-food ads (e.g., an automobile commercial) that lasted 30 seconds each and were recorded in the weeks prior to the study (the ads are listed in Table 2). Hence, ads that were in circulation at the time of the study were utilized. The six food and non-food ads were selected randomly by recording network television for 3 hours in the evening and selecting the first six food and non-food ads.
The reading material consisted of prose similar in topic and content to the TV programs. Specifically, the Internet Travel Planner (http://www.amazon.com/Internet-Travel-Planner-Trips-Online/dp/0762705795) by Michael Shapiro was used as reading material. Following the methods of McLaughlin (16), the material obtained a Flesch-Kincaid Grade Level of 7.3.
Participants were informed that they would be asked several questions about the material that they viewed or read during the TV and reading conditions. This procedure ensured that participants watched the TV programs and read the narrative, i.e., attended to and dedicated cognitive resources to these stimuli while eating.
Following the TV-ads condition, participants' memory for the ads from the TV program was quantified using methods similar to those of Halford (15). The memory task consisted of a list that contained descriptions of ads that were and were not in the TV program. As outlined in Table 2, 50% of the ads on the list were in the program and 50% were not. Participants were asked if they remembered each ad as being in the program. Ad memory was quantified by calculating the proportion of ads that participants remembered. Participants' familiarity with the ads was also quantified using similar methods, but this variable is not described in further detail because it was not of primary interest and did not correlate with the primary outcome variables, including energy intake. Ad memory data were fitted to a dichotomous Rasch model (17) to assess the reliability and construct validity of the ad recognition task and to test for differential item functioning (DIF) between males and females. Instrument reliability was acceptable, and DIF was not detected, indicating that item responses were not gender-biased. Average model fit was acceptable, although significant item-item interactions were identified at the level of individual respondents. In practice, such interactions equate with nonlinearities that threaten the validity of the person score. These effects were removed using the model error components, generating adjusted person scores that were then incorporated into all subsequent analyses.
The following assessments were completed during baseline.
The CPT-II is a norm-referenced computer-based test that measures an individual's ability to attend to and concentrate on prompts of visual stimuli. The CPT-II provides an objective measure of attention deficit/hyperactivity disorder (ADHD) symptoms and it has established reliability and validity (18, 19). The Confidence Index was used as an objective measure of distractibility, quantifying the extent to which a participant's responses reflect the performance of an ADHD population. Inattentiveness or the tendency to be easily distracted is a core feature of ADHD, and the majority of people (at least children and adolescents) diagnosed with ADHD exhibit this symptom (20).
Participants completed the WRAT-4, a norm-referenced measure of academic skills, to obtain an objective measure of reading level and to determine if the reading material utilized in the study was consistent with participants' reading level. Grade equivalents and standard scores are obtained with the WRAT-4.
The Eating Inventory is a 51-item self-report inventory that assesses dietary restraint (the intent to restrict energy intake), disinhibition (the tendency to overeat), and perceived hunger. The Eating Inventory has been found to be reliable and valid (21). In this study, the Eating Inventory was used to quantify dietary restraint, which is associated with body weight and energy intake (22, 23), and the relation between energy intake and emotional arousal induced by movie watching (10, 11).
Computerized VAS were completed before and after each test meal to measure subjective ratings of hunger, fullness, desire to eat, food craving, strength of cravings, and desire to eat something sweet, salty, and fatty. When completing the VAS, participants rated the intensity of these subjective states on a 100-unit line anchored from “not at all” to “extremely.” VAS have been found to have satisfactory reliability and validity (24).
Computerized VAMS were used to measure mood (happy vs. sad), alertness (alert vs. drowsy), tranquility (troubled vs. tranquil), anxiety (tense vs. relaxed), energy levels (lethargic vs. energetic), and calmness (calm vs. excited) before and after each test meal to determine if the TV programs or the reading material differentially affected mood. VAMS rely on the same 100-unit line as the VAS ratings and have been found to have satisfactory reliability and validity (25).
Prior to reporting to the clinic to eat breakfast on test days, participants were instructed to fast for 12 hours, refrain from vigorous exercise for 24 hours, and refrain from consuming alcohol for 48 hours. Upon arrival they completed a questionnaire to assess if they had fasted or were experiencing a cold or allergies. Participants reporting cold or allergy symptoms that affected their ability to taste or smell food were rescheduled. Female participants completed all tests meals during the luteal menstrual cycle phase.
At each test meal, participants were instructed to eat as much or as little as they wished. The length of time that participants remained in the room (32 minutes) was standardized across testing conditions. For the purpose of the study, an office at the PBRC was altered to accommodate the television and to provide a comfortable environment in which to eat. Book shelves and office equipment were removed or covered. Environmental stimuli in the room were identical across conditions (e.g., the TV was always in the room, but it was only on during the two TV conditions). Participants completed VAS and VAMS ratings before and after each test meal. Upon completion of the TV-ads condition, participants completed the memory task. Participants answered questions about the reading material after the reading condition.
The primary outcome variables were energy intake (kcal) and intake of kcal from fat, carbohydrate, and protein. Repeated measures analysis of variance (ANOVA) was used to test if energy intake differed significantly by condition (control, reading, TV-ads, TV-no ads). Sex was a between subjects factor.
Correlation analyses were used to examine the association of distractibility with energy intake and body weight, including sex as a partial covariate. Correlation analyses were also used to assess the relationship between memory for ads and energy intake and body weight. The distributions for ad memory were negatively skewed; therefore, non-parametric Spearman's rho correlation coefficients were reported for these variables. Memory for ads was measured during the TV-ads condition, but was correlated with energy intake in all conditions to determine if people with better memory for ads have greater energy intake in general, or if the associations only occur when certain stimuli, namely TV, are present.
The alpha level for the aforementioned primary outcome variables was 0.05. The number of correlations was limited to only those of primary interest and was planned a priori; therefore, alpha was set at 0.05 for the correlation analyses.
Repeated measures ANOVAs were used to test if change in subjective ratings of appetite and mood differed across conditions. Change scores (post-meal minus pre-meal) were calculated for the VAS and VAMS ratings. These change scores were subjected to repeated ANOVAs, which included condition (meal type) as the repeated factor and sex as a between subjects factor. Alpha was set at 0.01 for the VAS and VAMS analyses to help control alpha inflation.
To ensure data integrity, the presence of order effects was tested with repeated measures ANOVA, with contrasts to test for energy intake differences among the first, second, third, and fourth meal presented during the study. These tests were conducted with the variable of primary interest, energy intake.
The ANOVAs reported herein relied on the Greenhouse-Geisser p-value. All analyses were conducted with SPSS, Version 15, Software (Chicago, IL), with the exception of the Rasch model analysis, which were conducted using Winsteps, Version 3.65.1(Chicago, IL).
The demographic characteristics of the study sample are depicted in Table 3. The sample was 54% female and predominantly Caucasian. The mean dietary restraint score was 7.3, which indicated that the sample did not have high levels of dietary restraint (21). The mean BMI was 25.8 kg/m2, ranging from 20 to 35. The Word Reading and Sentence Comprehension Grade Equivalent scores from the WRAT-4 (12.0 and 11.8, respectively) indicated that the participants' reading level was above the reading level of the material in the reading condition (Flesch-Kincaid Grade Level of 7.3). Men scored higher on distractibility than women (Table 3), a finding that is consistent with disorders involving distractibility (e.g., attention-deficit/hyperactivity disorder) being more prevalent in males compared to females (26).
The repeated measures ANOVA indicated that energy intake did not vary by condition, F(2.8, 130.5) = 0.30, p = 0.81, and the condition by sex interaction was non-significant, F(2.8, 130.5) = 0.48, p = 0.69 (Figure 1). The effect sizes for these comparisons were very small (partial eta squared ≤ 0.01). Similar analyses indicated that macronutrient intake (kcal from fat, carbohydrate, and protein) did not differ significantly by condition (p-values > 0.65) or condition by sex (p-values > 0.29) (Figures 2a, 2b, and 2c). The analyses were also conducted using grams of food consumed as the dependent variables and the results were virtually identical.
The results of the correlation analyses are summarized in Table 4. The first row of results demonstrates that with sex entered as a partial covariate, distractibility was significantly associated with body weight, but not energy intake. Examination of the correlation coefficients for each sex indicated associations of moderate magnitude between distractibility and energy intake for men (r-values ranged from 0.20 to 0.36, p-values > 0.10, across testing conditions), but not women (r-values ranged from -0.15 to 0.07, p-values > 0.47). Conversely, distractibility was significantly associated with body weight for women (r = 0.46, p < 0.05), but not men (r = 0.23, p = 0.29). Memory for ads was significantly associated with body weight and with energy intake in the TV-no ads condition (p<0.05), and this relation was marginally significant in the TV-ads condition (p=0.06, Table 4).
Mean VAS and VAMS change scores (post-meal minus pre-meal) representing subjective levels of appetite and mood are depicted in Table 5. Change scores for VAS and VAMS ratings did not differ significantly (alpha = 0.01) by condition, as indicated in Table 5. The models also included effects for sex and sex by condition interactions, and these effects were not significant for VAS (p-values > 0.01) and VAMS (p-values > 0.03).
The test of order effects indicated that energy intake was significantly lower during the fourth (final) meal compared to the other meals (p-values < 0.01). The primary analyses were repeated with the fourth meal excluded and the results did not change. Therefore, we reported results from the analyses that included all data.
In contrast to previous studies, TVV was not found to influence energy intake. The failure of TVV to increase energy intake in this tightly controlled study is surprising since most (3-6), but not all (7), published studies found TVV to increase energy intake. The effect sizes reported in the literature vary widely; however, with TVV increasing energy intake by 11.5% (3) to 71% (5), and most study samples consisted of college-age females who scored high on dietary restraint. In such study samples, TVV might increase energy intake by disrupting dietary restraint, while this effect would not be observed in more representative samples of unrestrained adults. It is also possible that TVV affects energy intake through alteration of mood, which is consistent with earlier studies (10, 11). In our study, we controlled the effects of mood alteration on energy intake by selecting test stimuli (TV programs and reading material) that were mood neutral, and the VAMS data indicates that mood did not change differently among the testing conditions. Therefore, if mood alteration is necessary for TVV to affect energy intake, our study would not find an effect because change in mood did not differ among testing conditions.
A novel finding of the present study was that memory for ads was associated with energy intake, but this association was only significant when distracting stimuli (i.e., TV) were present. Specifically, the correlation between memory for ads and energy intake was significant in the TV-no ads condition and marginally significant in the TV-ads condition. Memory for ads was also significantly associated with body weight, suggesting that people with better memory for ads eat more food and/or are more sedentary, which would promote a positive energy balance and weight gain. After controlling for sex, distractibility was significantly associated with body weight but the correlation between distractibility and energy intake failed to reach statistical significance. Nevertheless, distractibility accounted for up to 13% of the variance in the energy intake of men during the TV-ads condition, and the significance of this correlation coefficient (r=0.36, p=0.11) was negatively affected by the sample size for males (n=22).
The findings from this study suggest that individual characteristics are associated with energy intake and body weight, and these characteristics might interact with environmental stimuli, e.g., TVV. Memory for ads was only associated with energy intake while viewing TV, and these data suggest that advertisements affect the energy intake of adults, which extends results found with children and adolescents (8, 9). The role of conditioning requires further examination since memory for ads was associated with energy intake in the TV-no ad condition, suggesting that the presence of TV, even if no ads are shown, contributes to the energy intake of people who have better memory for ads. It is possible that these individuals are more susceptible to conditioning and are more sensitive to TV as a cue for food intake. Importantly, the results from this study demonstrate that people can be identified who are more sensitive to the effect of environmental stimuli on energy intake and who would benefit from targeted interventions.
The measure of distractibility in this study, the Confidence Index of the CPT-II, reflects the degree to which participants' responses match those of people diagnosed with attention deficit hyperactivity disorder (ADHD). ADHD is associated with overweight and obesity (27, 28) and is more common on males (26). Distractibility accounted for 13% of the variance in the energy intake of men during the TV-ads condition. Consequently, distractibility in males is likely associated with energy intake, and the presence of distracting stimuli might increase the strength of this association. This would help explain the association between ADHD and obesity. This study did not a priori recruit participants who scored high or low on distractibility, and additional research is warranted to determine if TVV (or other tasks that require attention/cognitive resources) distracts susceptible individuals and results in passive over-consumption (12) or the disruption of habituation to food cues (13). Mood alteration during TVV might also interact with distractibility to modify energy intake, and this is worthy of future investigation.
The study reported herein has a number of strengths. First, the study included a control condition, a reading condition that served as a distracting task other than TVV, and TVV conditions that did and did not include ads. Second, the study sample's level of dietary restraint was low and did not correlate with energy intake (data not shown). Third, the study relied on a sample of adult men and women whose BMI ranged from the healthy (20 kg/m2) to obese (35 kg/m2) range, and the mean BMI (25.8 kg/m2) was similar to the national mean for the United States (29). Finally, the confounding effects of mood alteration on energy intake were controlled.
The study also has limitations. First, the study was conducted at a research center that offered exceptional experimental control, but the clinical environment is likely dissimilar to participants' natural eating environments. Nevertheless, significant efforts were made to make the eating environment pleasant and as comfortable as possible. Second, an order effect was detected; energy intake during the last food intake test was lower than the other food intake tests. The primary analyses were repeated with data from the last test excluded and the results were identical, as expected due to the random sequence of testing conditions.
In contrast to previous studies, TVV was not found to influence energy intake. Two novel findings from this study are: 1) memory for ads was associated with energy intake and body weight, and 2) distractibility was associated with body weight, and 13% of the variance in men's energy intake was accounted for by distractibility. The results suggest that individual characteristics are associated with energy intake and body weight, and these characteristics should be considered in food intake and intervention studies. The results also suggest that individual characteristics may play a significant role in the effect that environmental stimuli (i.e., TVV) have on energy intake, and further research is warranted to determine if these characteristic help explain energy intake mechanisms.
This work was partially supported by the National Institutes of Health (National Institute of Diabetes and Digestive and Kidney Diseases grant 1 K23 DK068052-01A2), and a CNRU Center Grant, 1P30 DK072476, entitled “Nutritional Programming: Environmental and Molecular Interactions” sponsored by the NIDDK. The content is solely the responsibility of the authors and does not necessarily represent the official view of the National Institutes of Health. Preliminary data from this study was presented at the Obesity Society 2007 Annual Meeting. The study design, execution, statistical analysis, data interpretation, and writing of the manuscript were the primary responsibilities of CKM, SMC, NM, and SDA. Medical monitoring, as well as writing the manuscript, was the responsibility of FLG. No authors have a conflict of interest.
Publisher's Disclaimer: “This is an un-copyedited author manuscript that has been accepted for publication in The American Journal of Clinical Nutrition, copyright American Society for Nutrition (ASN). This manuscript may not be duplicated or reproduced, other than for personal use or within the rule of ‘Fair Use of Copyrighted Materials’ (section 107, Title 17, US Code) without permission of the copyright owner, the ASN. The final copyedited article, which is the version of record, can be found at http://www.ajcn.org/. The ASN disclaims any responsibility or liability for errors or omissions in this version of the manuscript or in any version derived from it by the National Institutes of Health or other parties.”