- Email: [email protected]
Impact of moods and social context on eating behavior
Appetite (2001) 36, 111±118 doi:10.1006/appe.2000.0385, available online at http://www.idealibrary.com on
1
Original Article
Impact of moods and social context on eating behavior K.A. Patel and D.G. Schlundt Department of Psychology,Vanderbilt University (Received 30 October1999, revision 4 September 2000, accepted in revised form 20 October 2000, published electronically 26 January 2001)
The relationship of moods and social context to energy and nutrient intakes was examined to ascertain if these variables interact or function independently. The subjects were 78 predominantly white, obese women participating in weight-loss studies. Mean age was 367 (SD 76) and mean Body Mass Index was 321 (SD 36). Subjects completed 2-week baseline food diaries recording everything they ate, including moods and people present during the meals. Meals eaten in positive and negative moods were significantly larger than meals eaten in a neutral mood. Meals eaten with other people were significantly larger than meals eaten alone. There were no significant moods by social context interactions for total energy intake. Moods and social context functioned additively to increase the risk of over-eating. Macro nutrient analysis revealed only a main effect for social context. Percentage of calories from fat and protein were greater, whereas the percentage of carbohydrate was less in social context meals compared to meals eaten alone. Clinicians should conduct a functional analysis to assess exposure to the frequency and types of risky situations. Teaching people to cope more effectively with social situations and moods may increase the efficacy of weight loss and maintenance programs. # 2001 Academic Press
Introduction Being overweight has been implicated as a risk factor for a variety of health problems. Examining the factors that influence eating provides valuable information about risk and protective factors for over-eating. This information can be used to effectively facilitate and maintain weight loss. Food consumption increases in the presence of other people. The presence of other people altering individual behavior has been called social facilitation (Zajonc, 1965). The earliest work on the impact of the social context on food consumption used animals. Several studies of animal behavior demonstrated that animals eat more when in a group than when alone (Tolman & Wilson, 1965; Hoyenga & Aeschleman, 1969). Studies examining social facilitation of eating in humans have employed food dairies to evaluate Address for correspondence: Mr Kushal A. Patel, 301 Wilson Hall, Vanderbilt University, Nashville, TN 37240, U.S.A. 0195±6663/01/020111+08 $35.00/0
factors that facilitate or inhibit food consumption in natural settings. Food consumption increases in the presence of others, irrespective of whether the meal is eaten with friends, family or strangers (De Castro & Brewer, 1991; De Castro, 1994; Clendennen et al., 1994). Positive correlations between presence of other people and food consumption have been found for snacks, breakfast, lunch, and dinner meals (De Castro et al., 1990), for meals eaten at home and away from home (De Castro et al., 1990), and for meals eaten on weekdays and weekends and irrespective of whether alcohol was consumed (De Castro, 1991). Moods also influence food consumption. Studies examining participants in weight reduction programs have found that mild, moderate, and severely obese women increased their food intake in response to a variety of negative emotions such as anxiety, depression and anger (for reviews, see Ganley, 1989; Arnow et al., 1992; Steptoe et al., 1998). Over-eating in response to negative emotions has been found to occur in both obese and normal weight women. There is some evidence that obese subjects engage in more # 2001 Academic Press
112
K.A. Patel and D.G. Schlundt
emotional overeating than normal weight subjects (Foreyt et al., 1995; Schlundt & Zimering, 1988). Various studies have indicated that emotional eating is prevalent across various SES (socio-economic status) groups and has been found to be most frequent when people are alone, when the meal is supper or snack, and when the meal is eaten at home compared to away from home (Baumeister et al., 1994). Literature on disinhibited eating and dietary collapses has also provided information on the impact of negative moods on eating. Negative affect has been linked to disinhibition of eating. Dieters report that they often overeat in response to negative moods and obese women who experience weight fluctuations find it more difficult to control their eating in response to negative emotions (Eldredge et al., 1994). These findings have been confirmed by laboratory studies (Heatherton et al., 1998; Cools et al., 1992). The impact of positive mood on food intake has not been as well studied as the impact of negative mood on food intake. Studies that have looked at the relationship between positive mood and food intakes have found conflicting results. For example, some researchers have found no association between over-eating and positive moods (Schmitz, 1996; Davis et al., 1985; Lowe & Fisher, 1983) whereas others have found that positive mood is related to over-eating in social situations (Schlundt et al., 1988). One explanation for the failure to find a significant relationship between positive mood and food intake in some studies is that these studies did not examine this relationship in the context of the social situations in which these meals were eaten. The purpose of this study was to investigate the relationship between social context at meals and moods. In particular, we wanted to examine whether these factors interact or function independently. The first prediction of this study was that meals eaten in negative moods and positive mood should be larger than meals eaten in a neutral mood (absence of these mood states). The second prediction was that meals eaten in negative moods would be larger when they were eaten alone compared to when they were eaten in a social context. The opposite should be true for meals eaten in a positive mood. These meals should be larger when eaten in a social context compared to when they are eaten alone. The rationales for these predictions are grounded in the disinhibited eating literature which suggests that negative affect is one the main factors in disinhibited eating and that disinhibited eating is most frequent when a person is alone. Also, there is evidence in the mood and food intake literature that suggest that positive mood increases food intake in social situations.
Methods Subjects The subjects were 78 obese women who had participated in one of four weight-loss studies (Schlundt et al., 1993; Schlundt et al., 1992; Hill et al., 1989; Schlundt et al., 1991). The mean age for the subjects was 367 (SD 76) and their mean Body Mass Index was 321 (SD 36).
Procedure Obese subjects were obtained from pre-existing databases compiled from four weight management studies. These four studies had employed similar methods for recruiting subjects and collecting data. The generic method involved for recruited subjects was through newspaper notices. Subjects who responded to the newspaper notices were screened over the telephone to see if they met the weight inclusion criteria of being between 30% and 60% above their ideal weight based on the Metropolitan Life Tables (Metropolitan Life Insurance Company, 1983). Those who met these criteria were scheduled to attend two orientation sessions. During the first orientation session they were informed that they would have to maintain a 2-week diary in which they would record everything they ate during a 2-week period. The eating diaries were 52100 * 800 booklets containing 40 pages for recording information about their food intake. Each of the 40 pages contained a space for recording a single eating episode. Space was also provided for subjects to indicate whether the meal was a snack, breakfast, lunch or dinner. The subjects also recorded with whom they ate the meal (alone, friends, family or other), the place they ate the meal (home, work, restaurant, social event, car, other), and the time, date and day of the meal. Subjects checked any of 18 possible adjectives to indicate how they felt physically and emotionally during each eating episode. During the first orientation, subjects were trained to record the kinds of foods eaten, the amounts eaten and to code the foods eaten into six food groups (milk, meat, vegetable, fruit, bread and fat) based on the American Diabetic Association (ADA) exchange system. This training for subjects was provided by a nutritionist who gave feedback to subjects on whether they were accurately recording trial food entries. Subjects were also given two additional blank food diaries to ensure that they would have enough space to record food intakes over a 2-week period. Schlundt (1995) showed that results of food group coding corresponds quite well to the results of computerized nutrient analysis.
Mood, social context and eating 113
A second orientation meeting was conducted two weeks after the first orientation. The subject's food diaries were collected and a nutritionist was present to help subjects who had problems completing their food diaries.
computing the percentage of calories from protein, fat, and carbohydrates. Meals eaten with family or friends were grouped to form social context meals. Total energy intake and macronutrient intake for social context meals for each subject were calculated similarly to the alone meals.
Creating the mood variables Four mood variables were created. The four variables were Negative mood, Passive Negative mood, Positive mood and neutral moods. These mood variables were based on the Circumplex model of moods devised by Watson and Tellegen (1985). This 2-factor structure of mood has been shown to be both stable and robust. Neutral mood was coded as the absence of any mood for an eating episode. Therefore, neutral mood meals where those eating episodes for which a subject did not endorse any mood variable. These eating episodes were aggregated across meals for each subject and a mean was obtained. Hence, each subject contributed mean total energy intake and macronutrient intake (fat, carbohydrate and protein) for meals eaten in the absence of any mood. To remove the confounding effects of total calories, macronutrient values were adjusted to measure percentage of calories from protein, carbohydrate and fat. The mood adjectives excited and happy were combined to form the positive mood variable. Hence, if a subject indicated that she experienced even one of those mood adjectives for a particular eating episode, then she was coded as being in a positive mood for that eating episode. The mood adjectives, angry, nervous, stressed, upset and irritable were lumped together to form the negative mood variable. In a similar way, the mood variables of depressed, bored, tired and weak were combined to form the passive±negative mood variable. The negative mood variable was divided into passive± negative mood (low arousal) and active negative mood (high arousal). The rationale for this was to investigate if arousal levels for negative mood differentially impact food intakes. Previous research on negative mood and food intakes has tended to ignore levels of arousal in negative mood.
Modifications to the people variable The social context of meals was categorized into meals eaten alone and with other people. Total energy intakes for meals eaten alone for each subject was calculated by averaging calorie intake across all meals eaten alone by that subject. Macro nutrient intakes for meals eaten alone for each subject were calculated by averaging fat, protein and carbohydrate grams across all meals eaten alone for each subject and then
Data analysis The food items recorded in food diaries were entered in a computer program called SMAS (Self-Monitoring Analysis System). The SMAS program allowed the researcher to enter all the information associated with each eating episode like time and place of the eating episodes, moods during eating episodes, and nutrient information for each eating episode. There were more than 9000 eating episodes from the food diaries recorded in the SMAS file. The information from the SMAS files was processed and used to create an SPSS/PC file. Mean total energy and macronutrient intakes (percent calories from fat, protein and carbohydrate) for the various moods were calculated by aggregating nutrient information across eating episodes within subjects. Hence, each subject provided mean nutrient information for meals eaten in the presence of a positive mood, passive±negative mood, neutral mood and a negative mood. Each subject had at lest three eating episodes associated with these moods. The same steps were followed to calculate macronutrient intakes when alone and in a social situation. This aggregated information was the focus of the multi variate statistical analysis for this study.
Results Means and standard deviations for total energy intake and percentage of calories from protein, carbohydrate and fat by social context and mood are presented in Table 1.
Passive negative mood by social context Meals eaten in the presence of a passive±negative mood were larger than meals eaten in a neutral mood, F(1, 77) 621, p < 001. Passive negative moods had a relatively small effect size, explaining 7% of the variance in total energy intake. Social context meals were larger than meals eaten alone, F(1, 77) 7129, p < 0001. The effect size for social context was large, explaining 48% of the variance in total energy intake. Passive-negative mood and social context did not interact, F(1, 77) 01.
114
Nutrients
Total energy Kcalabcd Percentage of calories from fatdg Percentage of calories from proteind Percentage of calories from carbohydratede a
Positive mood
Negative mood
c
Neutral mood
Social mean SD
Alone mean SD
Social mean SD
Alone mean SD
Social mean SD
Alone mean SD
Social mean SD
Alone mean SD
6961 2568
5218 2590
6286 2075
5013 2349
60554 1593
4574 1628
5901 1774
4765 1789
44% 611
37% 612
41% 778
39% 712
42% 618
39% 612
42% 788
37% 712
17% 678
16% 556
18% 419
15% 741
18% 307
15% 524
17% 333
17% 467
39% 730
47% 714
41% 906
46% 914
40% 704
46% 713
41% 867
46% 812
Main effect for positive mood. Main effect for negative mood. Main effect for passive negative mood. d Main effect for social context during meals. e Interaction between negative mood and social context during meals. f Interaction between positive mood and social context during meals. g Interaction between passive negative mood and social context during meals. SD, standard deviation. b
Passive-negative mood
K.A. Patel and D.G. Schlundt
Table 1. Means and standard deviations for total energy intake and percentage of calories from protein, carbohydrate and fat by social context and moods
Mood, social context and eating 115
There was no significant difference between percentage of calories from fat (F(1, 77) 081), protein (F(1, 77) 135), and carbohydrates (F(1, 77) 012) for meals eaten in a passive±negative mood compared to neutral mood meals. There was a significant main effect for social context. Percentage of calories from fat (F(1, 77) 1958, p < 0001), and protein (F(1, 77) 1969, p < 0001) were significantly larger, whereas, percentage of calories from carbohydrate (F(1, 77) 3395, p < 0001) was significantly smaller in meals eaten in a social context compared to meals eaten alone. The only mood and social context interaction was for percentage of calories from carbohydrates, F(1, 77) 603, p < 001. Percentage of calories from carbohydrates at neutral and passive±negative meals were similar when these meals were eaten in a social context. However, for meals eaten alone, percentage of calories from carbohydrate was larger in neutral mood meals compared to passive±negative mood meals.
Positive mood by social context Meals eaten in a positive mood were larger than meals eaten in a neutral mood, F(1, 77) 1409, p < 0001. Positive moods had a moderate effect size, explaining 15% of the variance in total energy intake. Meals eaten in a social context were larger than meals eaten alone, F(1, 77) 4057, p < 0001. The effect size for social context was large, explaining 34% of the variance in total energy intake. Positive mood and social context during meals did not interact, F(1, 77) 025. There was no significant difference between percentage of calories from fat (F(1, 77) 192), protein (F(1, 77) 072), and carbohydrates (F(1, 77) 085) for meals eaten in a positive mood compared to neutral mood meals. There was a significant main effect for social context. Percentage of calories from fat (F(1, 77) 2040, p < 0001) was significantly larger whereas percentage of calories from carbohydrate (F(1, 77) 2560, p < 0001) was significantly smaller for meals eaten in a social context compared to meals eaten alone. Percentage of calories from protein did not differ significantly in meals eaten alone compared to those eaten in a social context (F(1, 77) 187). There were no significant interactions between positive mood and social context for percentage of calories from macronutrients in meals.
Negative mood by social context Meals eaten in a negative mood were larger than meals eaten in a neutral mood, F(1, 77) 389, p 005. Negative mood had a small effect size, explaining 5% of the variance in total energy intake. Social context
meals were larger than meals eaten alone, F(1, 77) 6439, p < 0001. The effect size for the social context was large, explaining 45% of the variance in total energy intake. Negative mood and social context during meals did not interact, F(1, 77) 050. There was no significant difference between percentage of calories from fat (F(1, 77) 244), protein (F(1, 77) 018), and carbohydrates (F(1, 77) 075) for negative mood meals compared to neutral mood meals. There was a significant main effect for social context. Percentage of calories from fat (F(1, 77) 740, p < 001) and protein (F(1, 77) 736, p < 001) were significantly larger whereas percentage of calories from carbohydrate (F(1, 77) 2099, p < 0001) was significantly smaller for meals eaten in a social context compared to those eaten alone. The only mood and social context interaction was for percentage of calories from fat, F(1, 77) 541, p < 005. Percentage of calories from fat was greater in social meals when they were eaten in a neutral mood compared to a negative mood. However, for meals eaten alone, percentage of calories from fat was larger for meals eaten in a negative mood compared to neutral mood.
Covariate analysis Various eating related variables were covaried from the total energy intake analysis to ascertain the robustness of the moods and social context effects on eating. The covaried variables included, alcohol consumption, subjects' age, resting metabolic rates, lean body weight in kilograms, body mass index, marital status, cognitive control/restraint of eating measured by the Restraint Scale (Herman & Mack, 1975), frequency of exercising, SCL-90 total symptom indexes measured by the SCL-90 (Derogatis et al., 1973), and frequency of binging measured by the Binge Eating Questionnaire (Halmi et al., 1981). Frequency of binging was the only variable that significantly affected the relationship of moods and social context to total energy intake. The effect of positive, negative and passive±negative moods on increasing total energy intake compared to neutral moods, disappeared when frequency of binging was accounted for. The effect size for social context during meals fell from an average effect size of approximately 43% (large) to an average effect size of approximately 7% (small). These results suggest that a large part of these mood and social context effects are contributed by individuals who have a tendency towards binge eating. In summary, total energy intakes were larger for meals eaten in a passive±negative mood, negative mood, or positive mood compared to meals eaten in a neutral mood. Positive mood had the largest effect of increasing
116
K.A. Patel and D.G. Schlundt
food intake among the three mood states. Also, total energy intake was higher for meals eaten in a social context compared to meals eaten alone irrespective of the mood state. There were no significant interactions between moods and social context for total energy intake at meals. Macronutrient analysis revealed that there was no main effect of moods for any of the macronutrients. Hence, there were no significant differences in percentage of calories from fat, protein, and carbohydrates for meals eaten in a neutral mood compared to meals eaten in either a negative mood, passive±negative mood or positive mood. The percentages of calories from fat and protein were greater, whereas the percentage of calories from carbohydrate was less in meals eaten in a social setting compared to meals eaten alone. The only social context and mood interactions for percentage of calories from macronutrients was for percentage of calories from fat for negative mood and percentage of calories from carbohydrates for passive±negative mood.
Discussion Only one of the study's hypotheses was supported. Total energy intake is larger for positive mood and negative moods (negative, passive±negative) compared to neutral mood. However, the hypothesized interaction between the different moods and the social situation was not supported. Moods and social context have an independent, additive impact on food intake. The results of this study support the findings in the social facilitation of food intake literature which has consistently illustrated that food intake increases when meals are eaten in a social context (De Castro, 1991; Clendenen et al., 1994; De Castro, 1994). Through mechanisms yet unknown, the presence of other people at meals dramatically increases a person's vulnerability to increased food intake. The impact of the social context (effect size around 50%) is large compared to the impact of moods (effect size around 9%). Clinicians should to pay close attention to the frequency of social meals for individuals exploring weight loss since social context is a strong antecedent factor for increased food intake. An interesting finding regarding percentage of calories from macronutrients in meals eaten in a social context versus meals eaten alone is that the percentage of calories from fat and protein were greater in meals eaten in a social setting while percentage of carbohydrates was less. These results may simply suggest that meals eaten with others differ in the kinds of foods eaten. For example, people may be more likely to consume high fat foods like ``steak'' which have been cooked with ample
fats (oils, butter) because the emphases is on taste. Also, eating high fat desert foods like cakes may be more common with meals eaten in a social setting. The results for the impact of passive±negative mood and negative mood on increasing food intake is consistent with the literature in this area. These negative moods do increase food intake but the impact is small as evident from the effect size (around 5% of the variance in total energy intake for negative mood and 7% for passive±negative mood). Clinicians working in the area of weight management should be cognizant that although these negative moods appears to have a small impact of increasing food intake, for some people, negative moods may be much more important factors in disrupting dietary adherence. This conclusion is supported by the covariance analysis showing that accounting for individual differences in binging frequency removes this effect. Also, unlike the social context effect which may be prevalent for most of the subjects, negative mood effects of increasing total energy intake may be present for a much smaller group of subjects. Hence, only a small subgroup of subjects show negative mood eating and this effect is diluted when these negative mood eaters are analysed as part of a larger group consisting of people who do not overeat in response to negative moods. There is a debate in the mood and food literature regarding the use of carbohydrates to elevate negative moods (Schlundt et al., 1993). Although the literature has conflicting results, a few studies have shown that people increase their carbohydrate consumption relative to other macronutrients (proteins and fat) in an attempt to self-medicate against negative mood (Christensen, 1993; Reid & Hammersley, 1995; Markus et al., 1998). Carbohydrates are thought to increase the level of serotonin in the brain by indirectly assisting the greater transfer of Tryptophan (serotonin precursor) across the blood±brain barrier. The results of this study revealed no preferential intake of carbohydrates during meals eaten in a negative mood. The results of positive mood on food intake are more difficult to interpret because of the conflicting information in the mood and food literature. As mentioned in the introduction, some studies have shown that positive mood increases food intake while others have revealed no such impact of positive mood. One possible explanation for this discrepancy is that the impact of positive mood on food intake may not be a stand alone effect. Positive mood may increase food intake only in social situations as demonstrated by Schlundt and Zimering (1988). This study's results suggest that the effect of positive mood on increasing food intake is both robust and independent from the social context. The dearth of information in this area has identified no mechanisms or
Mood, social context and eating 117
provided no theoretical framework for explaining and understanding this effect. The effect of positive mood on eating may be similar to negative mood in that it involves a disinhibition of eating control or positive mood may increase food intake via an associative learning mechanism where happiness has been associated with eating more food. This study has also illustrated that not only does positive mood increase food intake but it appears to have a stronger impact than negative moods. The effect size for positive mood on food intake is around 15% of the variance in energy intake which is substantially larger than effect size of 5% and 7% for negative mood and passive±negative mood respectively. This information is beneficial to clinicians who might spend considerable energy reducing a person's vulnerability to increased food intake by addressing only the negative moods, and ignoring positive mood. Clinicians need to understand the relative risk presented by both moods on increasing food intake. This study highlights the importance of adopting a functional analysis approach to eating behavior to get an accurate picture of the factors that impact a person's eating. Functional analysis focuses on the variables that influence eating with the goal of identifying the antecedents and consequences that influence and control eating behavior (Schlundt & Johnson, 1990). Through functional analysis, clinicians can get an accurate picture of a person's vulnerability to weight gain and likelihood of maintenance based on the high risk situations to which that person is exposed. For example, a person who has a stressful job and eats the majority of her meals in a social setting would be far more vulnerable to weight gain than a person who eats majority of these meals alone and is not stressed. The information gathered through functional analysis of eating behaviors has important treatment implications. Identifying the problem situations related to eating helps the clinician promote weight loss and maintenance by assisting patients to monitor and cope with these situations. In our example mentioned above, clinicians could use this knowledge to teach the person to cope with her stress before eating, hence reducing the probability of increasing food intake. Alternatively, clinicians could reduce the probability of increased food intake for the person by having her avoid these problem situations altogether, reducing her exposure to them, or helping her learn more effective ways to cope with them. There are some limitations to the generalizability of this study. All the subjects were women, predominately white and obese. This study cannot make the same claims for the impact of moods on food intake for males. Also, only obese women were included in this study,
hence, there is no information about the impact of mood on food intake for normal and underweight women. The results of this study should be viewed in the context of strengths and weaknesses of the food diary method of data collection employed in this study. Among the main criticisms for the diary data is that there are inaccuracies in reporting because the data is based on retrospective reporting. We tried to reduce inaccuracy due to retrospective reporting by instructing subjects to enter their moods in the diary before each meal and to enter meal content right after each meal was eaten. It was hoped that by having subjects enter their moods and meal content immediately after each meal would reduce errors associated with retrospective reporting of meals and moods. Another problem with this method of data collection is that different people may interpret the mood adjectives in a variety of ways. This may lead to inaccuracy in reporting due to idiosyncratic interpretations of the mood variables. We tried to reduce these errors in two ways. First, the subjects were provided with a variety of mood adjectives (28 adjectives) to ensure that they could find an adjective which was applicable to their mood state. Second, mood adjectives were grouped to form the negative and positive moods. This was undertaken to ensure that even though subjects had idiosyncratic ways of picking mood adjectives, by grouping adjectives we hoped that the general mood state (either positive or negative) for a particular meal was accurately captured. Another problem with dairy data is that it tends to underestimate total energy intake. This problem may not be as pertinent to this study considering that it was examining change in total energy and macronutrients within each subject based on moods and the social situation. In the analysis for this study, relative change of total energy and macronutrient intakes for each subject from one situation to the next is more important than the absolute means for energy and macronutrients. The food diary method of data collection was selected because we were interested in looking at eating behavior in a naturalistic setting. Measuring food intake in a laboratory setting would have not provided an adequate sampling of social events, moods, or food items. There are always trade-offs in internal and external validity when choosing laboratory or naturalistic settings for studying eating behavior.
References Arnow, G., Kenardy, J. & Agras, W. S. (1992). Binge eating among the obese: a descriptive study. Journal of Behavioral Medicine 15, 155±170.
118
K.A. Patel and D.G. Schlundt
Baumeister, R. F., Heatherton, T. F. & Tice, D. M. (1994). Losing control: how and why people fail at self-regulation. San Diego, CA: Academic Press, Inc. Christensen, L. (1993). Effects of eating behavior on mood: a review of the literature. International Journal of Eating Disorders 14, 171±183. Clendenen, V. I., Herman, P. C. & Polivy, J. (1994). Social facilitation of eating among friends and strangers. Appetite 23, 1±13. Cools, L., Schotte, D. E. & McNally, R. L. (1992). Emotional arousal and overeating in restrained eaters. Journal of Abnormal Psychology 1, 348±351. Davis, R., Freeman, R. J. & Solyom, L. (1985). Modd and food: an analysis of bulimic episodes. Journal of Psychiatric Research 19, 331±335. De Castro, J. M. (1991). Social facilitation of the spontaneous meal size of humans occurs on both weekdays and weekends. Physiology and Behavior 49, 1289±1291. De Castro, J. M. (1994). Family and friends produce greater social facilitation of food intake than other companions. Physiology and Behavior 56, 445±455. De Castro, J. M. & Brewer, E. M. (1991). The amount eaten in meals by humans is a power function of the number of people present. Physiology and Behavior 51, 121±125. De Castro, J. M., Brewer, E. M., Elmore, D. K. & Orozco, S. (1990). Social facilitation of the spontaneous meal size of humans occurs regardless of time, place alcohol or snacks. Appetite 15, 89±101. Derogatis, L. R., Lipman, R. S. & Covi, L. (1973). SCL-90: an outpatient psychiatric rating scale: preliminary report. Psychopoharmacology Bulletin 9, 13±27. Eldredge, K. L., Agras, W. S. & Arnow, B. (1994). The last supper: emotional determinants of pretreatment weight fluctuations in obese binge eaters. International Journal of Eating Disorders 16, 83±88. Foreyt, H. P., Brunner, R. L., Goodrick, G. K. & Cutter, G. (1995). Psychological correlates of weight fluctuations. International Journal of Eating Disorders 17, 263±275. Ganley, R. M. (1989). Emotions and eating in obesity: a review of the literature. International Journal of Eating Disorders 8, 343±361. Halmi, K. A., Falk, J. R. & Schwartz, E. (1981). Binge eating and vomiting: a survey of college populations. Psychological Medicine 11, 697±706. Heatherton, T. F., Striepe, M. & Wittenberg, L. (1998). Emotional distress and disinhibited eating: the role of self. Personality and Social Psychology Bulletin 24, 301±313. Herman, C. P. & Mack, D. (1975). Restrained and unrestrained eating. Journal of Abnormal Psychology 43, 647±660. Hill, J. O., Schlundt, D. G., Sbrocco, T., Sharp, T., PopeCordle, J., Stetson, B., Kaler, M. & Heim, C. (1989). Evaluation of an alternating-calorie diet with and without exercise in the treatment of obesity. American Journal of Clinical Nutrition 50, 248±254.
Hoyenga, K. T. & Aeschleman, S. (1969). Social facilitation of eating in the rat. Psychonomic Science 14, 239±241. Lowe, M. R. & Fisher, E. B. (1983). Emotional reactivity, emotional eating, and obesity: a naturalistic study. Journal of Behavioral Medicine 6, 135±149. Markus, C. R., Panhuysen, G., Tuiten, A., Koppeschaar, H., Fekkes, D. & Peters, M. L. (1998). Does Carbohydraterich, protein-poor food prevent a deterioration of mood and cognitive performance of stress-prone subjects when subjected to a stressful task? Appetite 32, 49±65. Metropolitan Life Insurance Company (1984). Height and weight tables. Statistical Bulletin Metropolitan Insurance Company 64, 2±9. Reid, M. & Hammersley, R. (1995). Effects of carbohydrate intake on subsequent food intake and mood states. Physiology and Behavior 58, 421±427. Schlundt, D. G. (1995). Assessment of Specific Eating Behaviors and Eating Style. In D. Allision (Ed.), Handbook of assessment methods for eating behaviors and weightrelated problems: measures, theory and research. Pp. 241±302. Thousands Oaks, London & New Delhi: Sage Publications. Schlundt, D. G. & Johnson, W. G. (1990). Eating disorders: assessment and treatment. Boston, MA: Allyn & Bacon. Schlundt, D. G. & Zimering, R. T. (1988). The dieter's inventory of eating temptations: a measure of weight control competence. Addictive Behaviors 13, 151±164. Schlundt, D. G., Sbrocco, T. & Bell, C. (1988). Identification of high-risk situations in a behavioral weight loss program: application of the relapse prevention model. International Journal of Obesity 13, 223±234. Schlundt, D. G., Taylor, D., Hill, J. O., Sbrocco, T., PopeCordle, J., Kasser, T. & Arnold, D. (1991). A behavioral taxonomy of obese female participants in a weight-loss program. American Journal of Clinical Nutrition 53, 1151±1158. Schlundt, D. G., Hill, J. O., Pope-Cordle, J. & Sharp, T. (1992). The role of breakfast in the treatment of obesity: a randomized clinical trial. American Journal of Clinical Nutrition 55, 645±651. Schlundt, D. G., Virts, K. L., Sbrocco, T. & Pope-Cordle, J. (1993). A sequential behavioral analysis of craving sweets in obese women. Addictive Behavior 18, 67±80. Schmitz, B. A. (1996). The relationship between affect and binge eating. Dissertation Abstracts International: Section B: The Sciences and Engineering 56, 7055. Steptoe, A., Lipsey, Z. & Wardle, J. (1998). Stress, hassles and variations in alcohol consumption, food choice and physical exercise: a diary study. British Journal of Health Psychology 3, 51±63. Tolman, C. W. & Wilson, G. F. (1965). Social feeding in domestic chicks. Animal Behavior 13, 134±142. Watson, D. & Tellegen, A. (1985). Toward a consensual structure of mood. Psychological Bulletin 98, 219±235. Zajonc, R. B. (1965). Social facilitation. Science 149, 269±274.
1
Original Article
Impact of moods and social context on eating behavior K.A. Patel and D.G. Schlundt Department of Psychology,Vanderbilt University (Received 30 October1999, revision 4 September 2000, accepted in revised form 20 October 2000, published electronically 26 January 2001)
The relationship of moods and social context to energy and nutrient intakes was examined to ascertain if these variables interact or function independently. The subjects were 78 predominantly white, obese women participating in weight-loss studies. Mean age was 367 (SD 76) and mean Body Mass Index was 321 (SD 36). Subjects completed 2-week baseline food diaries recording everything they ate, including moods and people present during the meals. Meals eaten in positive and negative moods were significantly larger than meals eaten in a neutral mood. Meals eaten with other people were significantly larger than meals eaten alone. There were no significant moods by social context interactions for total energy intake. Moods and social context functioned additively to increase the risk of over-eating. Macro nutrient analysis revealed only a main effect for social context. Percentage of calories from fat and protein were greater, whereas the percentage of carbohydrate was less in social context meals compared to meals eaten alone. Clinicians should conduct a functional analysis to assess exposure to the frequency and types of risky situations. Teaching people to cope more effectively with social situations and moods may increase the efficacy of weight loss and maintenance programs. # 2001 Academic Press
Introduction Being overweight has been implicated as a risk factor for a variety of health problems. Examining the factors that influence eating provides valuable information about risk and protective factors for over-eating. This information can be used to effectively facilitate and maintain weight loss. Food consumption increases in the presence of other people. The presence of other people altering individual behavior has been called social facilitation (Zajonc, 1965). The earliest work on the impact of the social context on food consumption used animals. Several studies of animal behavior demonstrated that animals eat more when in a group than when alone (Tolman & Wilson, 1965; Hoyenga & Aeschleman, 1969). Studies examining social facilitation of eating in humans have employed food dairies to evaluate Address for correspondence: Mr Kushal A. Patel, 301 Wilson Hall, Vanderbilt University, Nashville, TN 37240, U.S.A. 0195±6663/01/020111+08 $35.00/0
factors that facilitate or inhibit food consumption in natural settings. Food consumption increases in the presence of others, irrespective of whether the meal is eaten with friends, family or strangers (De Castro & Brewer, 1991; De Castro, 1994; Clendennen et al., 1994). Positive correlations between presence of other people and food consumption have been found for snacks, breakfast, lunch, and dinner meals (De Castro et al., 1990), for meals eaten at home and away from home (De Castro et al., 1990), and for meals eaten on weekdays and weekends and irrespective of whether alcohol was consumed (De Castro, 1991). Moods also influence food consumption. Studies examining participants in weight reduction programs have found that mild, moderate, and severely obese women increased their food intake in response to a variety of negative emotions such as anxiety, depression and anger (for reviews, see Ganley, 1989; Arnow et al., 1992; Steptoe et al., 1998). Over-eating in response to negative emotions has been found to occur in both obese and normal weight women. There is some evidence that obese subjects engage in more # 2001 Academic Press
112
K.A. Patel and D.G. Schlundt
emotional overeating than normal weight subjects (Foreyt et al., 1995; Schlundt & Zimering, 1988). Various studies have indicated that emotional eating is prevalent across various SES (socio-economic status) groups and has been found to be most frequent when people are alone, when the meal is supper or snack, and when the meal is eaten at home compared to away from home (Baumeister et al., 1994). Literature on disinhibited eating and dietary collapses has also provided information on the impact of negative moods on eating. Negative affect has been linked to disinhibition of eating. Dieters report that they often overeat in response to negative moods and obese women who experience weight fluctuations find it more difficult to control their eating in response to negative emotions (Eldredge et al., 1994). These findings have been confirmed by laboratory studies (Heatherton et al., 1998; Cools et al., 1992). The impact of positive mood on food intake has not been as well studied as the impact of negative mood on food intake. Studies that have looked at the relationship between positive mood and food intakes have found conflicting results. For example, some researchers have found no association between over-eating and positive moods (Schmitz, 1996; Davis et al., 1985; Lowe & Fisher, 1983) whereas others have found that positive mood is related to over-eating in social situations (Schlundt et al., 1988). One explanation for the failure to find a significant relationship between positive mood and food intake in some studies is that these studies did not examine this relationship in the context of the social situations in which these meals were eaten. The purpose of this study was to investigate the relationship between social context at meals and moods. In particular, we wanted to examine whether these factors interact or function independently. The first prediction of this study was that meals eaten in negative moods and positive mood should be larger than meals eaten in a neutral mood (absence of these mood states). The second prediction was that meals eaten in negative moods would be larger when they were eaten alone compared to when they were eaten in a social context. The opposite should be true for meals eaten in a positive mood. These meals should be larger when eaten in a social context compared to when they are eaten alone. The rationales for these predictions are grounded in the disinhibited eating literature which suggests that negative affect is one the main factors in disinhibited eating and that disinhibited eating is most frequent when a person is alone. Also, there is evidence in the mood and food intake literature that suggest that positive mood increases food intake in social situations.
Methods Subjects The subjects were 78 obese women who had participated in one of four weight-loss studies (Schlundt et al., 1993; Schlundt et al., 1992; Hill et al., 1989; Schlundt et al., 1991). The mean age for the subjects was 367 (SD 76) and their mean Body Mass Index was 321 (SD 36).
Procedure Obese subjects were obtained from pre-existing databases compiled from four weight management studies. These four studies had employed similar methods for recruiting subjects and collecting data. The generic method involved for recruited subjects was through newspaper notices. Subjects who responded to the newspaper notices were screened over the telephone to see if they met the weight inclusion criteria of being between 30% and 60% above their ideal weight based on the Metropolitan Life Tables (Metropolitan Life Insurance Company, 1983). Those who met these criteria were scheduled to attend two orientation sessions. During the first orientation session they were informed that they would have to maintain a 2-week diary in which they would record everything they ate during a 2-week period. The eating diaries were 52100 * 800 booklets containing 40 pages for recording information about their food intake. Each of the 40 pages contained a space for recording a single eating episode. Space was also provided for subjects to indicate whether the meal was a snack, breakfast, lunch or dinner. The subjects also recorded with whom they ate the meal (alone, friends, family or other), the place they ate the meal (home, work, restaurant, social event, car, other), and the time, date and day of the meal. Subjects checked any of 18 possible adjectives to indicate how they felt physically and emotionally during each eating episode. During the first orientation, subjects were trained to record the kinds of foods eaten, the amounts eaten and to code the foods eaten into six food groups (milk, meat, vegetable, fruit, bread and fat) based on the American Diabetic Association (ADA) exchange system. This training for subjects was provided by a nutritionist who gave feedback to subjects on whether they were accurately recording trial food entries. Subjects were also given two additional blank food diaries to ensure that they would have enough space to record food intakes over a 2-week period. Schlundt (1995) showed that results of food group coding corresponds quite well to the results of computerized nutrient analysis.
Mood, social context and eating 113
A second orientation meeting was conducted two weeks after the first orientation. The subject's food diaries were collected and a nutritionist was present to help subjects who had problems completing their food diaries.
computing the percentage of calories from protein, fat, and carbohydrates. Meals eaten with family or friends were grouped to form social context meals. Total energy intake and macronutrient intake for social context meals for each subject were calculated similarly to the alone meals.
Creating the mood variables Four mood variables were created. The four variables were Negative mood, Passive Negative mood, Positive mood and neutral moods. These mood variables were based on the Circumplex model of moods devised by Watson and Tellegen (1985). This 2-factor structure of mood has been shown to be both stable and robust. Neutral mood was coded as the absence of any mood for an eating episode. Therefore, neutral mood meals where those eating episodes for which a subject did not endorse any mood variable. These eating episodes were aggregated across meals for each subject and a mean was obtained. Hence, each subject contributed mean total energy intake and macronutrient intake (fat, carbohydrate and protein) for meals eaten in the absence of any mood. To remove the confounding effects of total calories, macronutrient values were adjusted to measure percentage of calories from protein, carbohydrate and fat. The mood adjectives excited and happy were combined to form the positive mood variable. Hence, if a subject indicated that she experienced even one of those mood adjectives for a particular eating episode, then she was coded as being in a positive mood for that eating episode. The mood adjectives, angry, nervous, stressed, upset and irritable were lumped together to form the negative mood variable. In a similar way, the mood variables of depressed, bored, tired and weak were combined to form the passive±negative mood variable. The negative mood variable was divided into passive± negative mood (low arousal) and active negative mood (high arousal). The rationale for this was to investigate if arousal levels for negative mood differentially impact food intakes. Previous research on negative mood and food intakes has tended to ignore levels of arousal in negative mood.
Modifications to the people variable The social context of meals was categorized into meals eaten alone and with other people. Total energy intakes for meals eaten alone for each subject was calculated by averaging calorie intake across all meals eaten alone by that subject. Macro nutrient intakes for meals eaten alone for each subject were calculated by averaging fat, protein and carbohydrate grams across all meals eaten alone for each subject and then
Data analysis The food items recorded in food diaries were entered in a computer program called SMAS (Self-Monitoring Analysis System). The SMAS program allowed the researcher to enter all the information associated with each eating episode like time and place of the eating episodes, moods during eating episodes, and nutrient information for each eating episode. There were more than 9000 eating episodes from the food diaries recorded in the SMAS file. The information from the SMAS files was processed and used to create an SPSS/PC file. Mean total energy and macronutrient intakes (percent calories from fat, protein and carbohydrate) for the various moods were calculated by aggregating nutrient information across eating episodes within subjects. Hence, each subject provided mean nutrient information for meals eaten in the presence of a positive mood, passive±negative mood, neutral mood and a negative mood. Each subject had at lest three eating episodes associated with these moods. The same steps were followed to calculate macronutrient intakes when alone and in a social situation. This aggregated information was the focus of the multi variate statistical analysis for this study.
Results Means and standard deviations for total energy intake and percentage of calories from protein, carbohydrate and fat by social context and mood are presented in Table 1.
Passive negative mood by social context Meals eaten in the presence of a passive±negative mood were larger than meals eaten in a neutral mood, F(1, 77) 621, p < 001. Passive negative moods had a relatively small effect size, explaining 7% of the variance in total energy intake. Social context meals were larger than meals eaten alone, F(1, 77) 7129, p < 0001. The effect size for social context was large, explaining 48% of the variance in total energy intake. Passive-negative mood and social context did not interact, F(1, 77) 01.
114
Nutrients
Total energy Kcalabcd Percentage of calories from fatdg Percentage of calories from proteind Percentage of calories from carbohydratede a
Positive mood
Negative mood
c
Neutral mood
Social mean SD
Alone mean SD
Social mean SD
Alone mean SD
Social mean SD
Alone mean SD
Social mean SD
Alone mean SD
6961 2568
5218 2590
6286 2075
5013 2349
60554 1593
4574 1628
5901 1774
4765 1789
44% 611
37% 612
41% 778
39% 712
42% 618
39% 612
42% 788
37% 712
17% 678
16% 556
18% 419
15% 741
18% 307
15% 524
17% 333
17% 467
39% 730
47% 714
41% 906
46% 914
40% 704
46% 713
41% 867
46% 812
Main effect for positive mood. Main effect for negative mood. Main effect for passive negative mood. d Main effect for social context during meals. e Interaction between negative mood and social context during meals. f Interaction between positive mood and social context during meals. g Interaction between passive negative mood and social context during meals. SD, standard deviation. b
Passive-negative mood
K.A. Patel and D.G. Schlundt
Table 1. Means and standard deviations for total energy intake and percentage of calories from protein, carbohydrate and fat by social context and moods
Mood, social context and eating 115
There was no significant difference between percentage of calories from fat (F(1, 77) 081), protein (F(1, 77) 135), and carbohydrates (F(1, 77) 012) for meals eaten in a passive±negative mood compared to neutral mood meals. There was a significant main effect for social context. Percentage of calories from fat (F(1, 77) 1958, p < 0001), and protein (F(1, 77) 1969, p < 0001) were significantly larger, whereas, percentage of calories from carbohydrate (F(1, 77) 3395, p < 0001) was significantly smaller in meals eaten in a social context compared to meals eaten alone. The only mood and social context interaction was for percentage of calories from carbohydrates, F(1, 77) 603, p < 001. Percentage of calories from carbohydrates at neutral and passive±negative meals were similar when these meals were eaten in a social context. However, for meals eaten alone, percentage of calories from carbohydrate was larger in neutral mood meals compared to passive±negative mood meals.
Positive mood by social context Meals eaten in a positive mood were larger than meals eaten in a neutral mood, F(1, 77) 1409, p < 0001. Positive moods had a moderate effect size, explaining 15% of the variance in total energy intake. Meals eaten in a social context were larger than meals eaten alone, F(1, 77) 4057, p < 0001. The effect size for social context was large, explaining 34% of the variance in total energy intake. Positive mood and social context during meals did not interact, F(1, 77) 025. There was no significant difference between percentage of calories from fat (F(1, 77) 192), protein (F(1, 77) 072), and carbohydrates (F(1, 77) 085) for meals eaten in a positive mood compared to neutral mood meals. There was a significant main effect for social context. Percentage of calories from fat (F(1, 77) 2040, p < 0001) was significantly larger whereas percentage of calories from carbohydrate (F(1, 77) 2560, p < 0001) was significantly smaller for meals eaten in a social context compared to meals eaten alone. Percentage of calories from protein did not differ significantly in meals eaten alone compared to those eaten in a social context (F(1, 77) 187). There were no significant interactions between positive mood and social context for percentage of calories from macronutrients in meals.
Negative mood by social context Meals eaten in a negative mood were larger than meals eaten in a neutral mood, F(1, 77) 389, p 005. Negative mood had a small effect size, explaining 5% of the variance in total energy intake. Social context
meals were larger than meals eaten alone, F(1, 77) 6439, p < 0001. The effect size for the social context was large, explaining 45% of the variance in total energy intake. Negative mood and social context during meals did not interact, F(1, 77) 050. There was no significant difference between percentage of calories from fat (F(1, 77) 244), protein (F(1, 77) 018), and carbohydrates (F(1, 77) 075) for negative mood meals compared to neutral mood meals. There was a significant main effect for social context. Percentage of calories from fat (F(1, 77) 740, p < 001) and protein (F(1, 77) 736, p < 001) were significantly larger whereas percentage of calories from carbohydrate (F(1, 77) 2099, p < 0001) was significantly smaller for meals eaten in a social context compared to those eaten alone. The only mood and social context interaction was for percentage of calories from fat, F(1, 77) 541, p < 005. Percentage of calories from fat was greater in social meals when they were eaten in a neutral mood compared to a negative mood. However, for meals eaten alone, percentage of calories from fat was larger for meals eaten in a negative mood compared to neutral mood.
Covariate analysis Various eating related variables were covaried from the total energy intake analysis to ascertain the robustness of the moods and social context effects on eating. The covaried variables included, alcohol consumption, subjects' age, resting metabolic rates, lean body weight in kilograms, body mass index, marital status, cognitive control/restraint of eating measured by the Restraint Scale (Herman & Mack, 1975), frequency of exercising, SCL-90 total symptom indexes measured by the SCL-90 (Derogatis et al., 1973), and frequency of binging measured by the Binge Eating Questionnaire (Halmi et al., 1981). Frequency of binging was the only variable that significantly affected the relationship of moods and social context to total energy intake. The effect of positive, negative and passive±negative moods on increasing total energy intake compared to neutral moods, disappeared when frequency of binging was accounted for. The effect size for social context during meals fell from an average effect size of approximately 43% (large) to an average effect size of approximately 7% (small). These results suggest that a large part of these mood and social context effects are contributed by individuals who have a tendency towards binge eating. In summary, total energy intakes were larger for meals eaten in a passive±negative mood, negative mood, or positive mood compared to meals eaten in a neutral mood. Positive mood had the largest effect of increasing
116
K.A. Patel and D.G. Schlundt
food intake among the three mood states. Also, total energy intake was higher for meals eaten in a social context compared to meals eaten alone irrespective of the mood state. There were no significant interactions between moods and social context for total energy intake at meals. Macronutrient analysis revealed that there was no main effect of moods for any of the macronutrients. Hence, there were no significant differences in percentage of calories from fat, protein, and carbohydrates for meals eaten in a neutral mood compared to meals eaten in either a negative mood, passive±negative mood or positive mood. The percentages of calories from fat and protein were greater, whereas the percentage of calories from carbohydrate was less in meals eaten in a social setting compared to meals eaten alone. The only social context and mood interactions for percentage of calories from macronutrients was for percentage of calories from fat for negative mood and percentage of calories from carbohydrates for passive±negative mood.
Discussion Only one of the study's hypotheses was supported. Total energy intake is larger for positive mood and negative moods (negative, passive±negative) compared to neutral mood. However, the hypothesized interaction between the different moods and the social situation was not supported. Moods and social context have an independent, additive impact on food intake. The results of this study support the findings in the social facilitation of food intake literature which has consistently illustrated that food intake increases when meals are eaten in a social context (De Castro, 1991; Clendenen et al., 1994; De Castro, 1994). Through mechanisms yet unknown, the presence of other people at meals dramatically increases a person's vulnerability to increased food intake. The impact of the social context (effect size around 50%) is large compared to the impact of moods (effect size around 9%). Clinicians should to pay close attention to the frequency of social meals for individuals exploring weight loss since social context is a strong antecedent factor for increased food intake. An interesting finding regarding percentage of calories from macronutrients in meals eaten in a social context versus meals eaten alone is that the percentage of calories from fat and protein were greater in meals eaten in a social setting while percentage of carbohydrates was less. These results may simply suggest that meals eaten with others differ in the kinds of foods eaten. For example, people may be more likely to consume high fat foods like ``steak'' which have been cooked with ample
fats (oils, butter) because the emphases is on taste. Also, eating high fat desert foods like cakes may be more common with meals eaten in a social setting. The results for the impact of passive±negative mood and negative mood on increasing food intake is consistent with the literature in this area. These negative moods do increase food intake but the impact is small as evident from the effect size (around 5% of the variance in total energy intake for negative mood and 7% for passive±negative mood). Clinicians working in the area of weight management should be cognizant that although these negative moods appears to have a small impact of increasing food intake, for some people, negative moods may be much more important factors in disrupting dietary adherence. This conclusion is supported by the covariance analysis showing that accounting for individual differences in binging frequency removes this effect. Also, unlike the social context effect which may be prevalent for most of the subjects, negative mood effects of increasing total energy intake may be present for a much smaller group of subjects. Hence, only a small subgroup of subjects show negative mood eating and this effect is diluted when these negative mood eaters are analysed as part of a larger group consisting of people who do not overeat in response to negative moods. There is a debate in the mood and food literature regarding the use of carbohydrates to elevate negative moods (Schlundt et al., 1993). Although the literature has conflicting results, a few studies have shown that people increase their carbohydrate consumption relative to other macronutrients (proteins and fat) in an attempt to self-medicate against negative mood (Christensen, 1993; Reid & Hammersley, 1995; Markus et al., 1998). Carbohydrates are thought to increase the level of serotonin in the brain by indirectly assisting the greater transfer of Tryptophan (serotonin precursor) across the blood±brain barrier. The results of this study revealed no preferential intake of carbohydrates during meals eaten in a negative mood. The results of positive mood on food intake are more difficult to interpret because of the conflicting information in the mood and food literature. As mentioned in the introduction, some studies have shown that positive mood increases food intake while others have revealed no such impact of positive mood. One possible explanation for this discrepancy is that the impact of positive mood on food intake may not be a stand alone effect. Positive mood may increase food intake only in social situations as demonstrated by Schlundt and Zimering (1988). This study's results suggest that the effect of positive mood on increasing food intake is both robust and independent from the social context. The dearth of information in this area has identified no mechanisms or
Mood, social context and eating 117
provided no theoretical framework for explaining and understanding this effect. The effect of positive mood on eating may be similar to negative mood in that it involves a disinhibition of eating control or positive mood may increase food intake via an associative learning mechanism where happiness has been associated with eating more food. This study has also illustrated that not only does positive mood increase food intake but it appears to have a stronger impact than negative moods. The effect size for positive mood on food intake is around 15% of the variance in energy intake which is substantially larger than effect size of 5% and 7% for negative mood and passive±negative mood respectively. This information is beneficial to clinicians who might spend considerable energy reducing a person's vulnerability to increased food intake by addressing only the negative moods, and ignoring positive mood. Clinicians need to understand the relative risk presented by both moods on increasing food intake. This study highlights the importance of adopting a functional analysis approach to eating behavior to get an accurate picture of the factors that impact a person's eating. Functional analysis focuses on the variables that influence eating with the goal of identifying the antecedents and consequences that influence and control eating behavior (Schlundt & Johnson, 1990). Through functional analysis, clinicians can get an accurate picture of a person's vulnerability to weight gain and likelihood of maintenance based on the high risk situations to which that person is exposed. For example, a person who has a stressful job and eats the majority of her meals in a social setting would be far more vulnerable to weight gain than a person who eats majority of these meals alone and is not stressed. The information gathered through functional analysis of eating behaviors has important treatment implications. Identifying the problem situations related to eating helps the clinician promote weight loss and maintenance by assisting patients to monitor and cope with these situations. In our example mentioned above, clinicians could use this knowledge to teach the person to cope with her stress before eating, hence reducing the probability of increasing food intake. Alternatively, clinicians could reduce the probability of increased food intake for the person by having her avoid these problem situations altogether, reducing her exposure to them, or helping her learn more effective ways to cope with them. There are some limitations to the generalizability of this study. All the subjects were women, predominately white and obese. This study cannot make the same claims for the impact of moods on food intake for males. Also, only obese women were included in this study,
hence, there is no information about the impact of mood on food intake for normal and underweight women. The results of this study should be viewed in the context of strengths and weaknesses of the food diary method of data collection employed in this study. Among the main criticisms for the diary data is that there are inaccuracies in reporting because the data is based on retrospective reporting. We tried to reduce inaccuracy due to retrospective reporting by instructing subjects to enter their moods in the diary before each meal and to enter meal content right after each meal was eaten. It was hoped that by having subjects enter their moods and meal content immediately after each meal would reduce errors associated with retrospective reporting of meals and moods. Another problem with this method of data collection is that different people may interpret the mood adjectives in a variety of ways. This may lead to inaccuracy in reporting due to idiosyncratic interpretations of the mood variables. We tried to reduce these errors in two ways. First, the subjects were provided with a variety of mood adjectives (28 adjectives) to ensure that they could find an adjective which was applicable to their mood state. Second, mood adjectives were grouped to form the negative and positive moods. This was undertaken to ensure that even though subjects had idiosyncratic ways of picking mood adjectives, by grouping adjectives we hoped that the general mood state (either positive or negative) for a particular meal was accurately captured. Another problem with dairy data is that it tends to underestimate total energy intake. This problem may not be as pertinent to this study considering that it was examining change in total energy and macronutrients within each subject based on moods and the social situation. In the analysis for this study, relative change of total energy and macronutrient intakes for each subject from one situation to the next is more important than the absolute means for energy and macronutrients. The food diary method of data collection was selected because we were interested in looking at eating behavior in a naturalistic setting. Measuring food intake in a laboratory setting would have not provided an adequate sampling of social events, moods, or food items. There are always trade-offs in internal and external validity when choosing laboratory or naturalistic settings for studying eating behavior.
References Arnow, G., Kenardy, J. & Agras, W. S. (1992). Binge eating among the obese: a descriptive study. Journal of Behavioral Medicine 15, 155±170.
118
K.A. Patel and D.G. Schlundt
Baumeister, R. F., Heatherton, T. F. & Tice, D. M. (1994). Losing control: how and why people fail at self-regulation. San Diego, CA: Academic Press, Inc. Christensen, L. (1993). Effects of eating behavior on mood: a review of the literature. International Journal of Eating Disorders 14, 171±183. Clendenen, V. I., Herman, P. C. & Polivy, J. (1994). Social facilitation of eating among friends and strangers. Appetite 23, 1±13. Cools, L., Schotte, D. E. & McNally, R. L. (1992). Emotional arousal and overeating in restrained eaters. Journal of Abnormal Psychology 1, 348±351. Davis, R., Freeman, R. J. & Solyom, L. (1985). Modd and food: an analysis of bulimic episodes. Journal of Psychiatric Research 19, 331±335. De Castro, J. M. (1991). Social facilitation of the spontaneous meal size of humans occurs on both weekdays and weekends. Physiology and Behavior 49, 1289±1291. De Castro, J. M. (1994). Family and friends produce greater social facilitation of food intake than other companions. Physiology and Behavior 56, 445±455. De Castro, J. M. & Brewer, E. M. (1991). The amount eaten in meals by humans is a power function of the number of people present. Physiology and Behavior 51, 121±125. De Castro, J. M., Brewer, E. M., Elmore, D. K. & Orozco, S. (1990). Social facilitation of the spontaneous meal size of humans occurs regardless of time, place alcohol or snacks. Appetite 15, 89±101. Derogatis, L. R., Lipman, R. S. & Covi, L. (1973). SCL-90: an outpatient psychiatric rating scale: preliminary report. Psychopoharmacology Bulletin 9, 13±27. Eldredge, K. L., Agras, W. S. & Arnow, B. (1994). The last supper: emotional determinants of pretreatment weight fluctuations in obese binge eaters. International Journal of Eating Disorders 16, 83±88. Foreyt, H. P., Brunner, R. L., Goodrick, G. K. & Cutter, G. (1995). Psychological correlates of weight fluctuations. International Journal of Eating Disorders 17, 263±275. Ganley, R. M. (1989). Emotions and eating in obesity: a review of the literature. International Journal of Eating Disorders 8, 343±361. Halmi, K. A., Falk, J. R. & Schwartz, E. (1981). Binge eating and vomiting: a survey of college populations. Psychological Medicine 11, 697±706. Heatherton, T. F., Striepe, M. & Wittenberg, L. (1998). Emotional distress and disinhibited eating: the role of self. Personality and Social Psychology Bulletin 24, 301±313. Herman, C. P. & Mack, D. (1975). Restrained and unrestrained eating. Journal of Abnormal Psychology 43, 647±660. Hill, J. O., Schlundt, D. G., Sbrocco, T., Sharp, T., PopeCordle, J., Stetson, B., Kaler, M. & Heim, C. (1989). Evaluation of an alternating-calorie diet with and without exercise in the treatment of obesity. American Journal of Clinical Nutrition 50, 248±254.
Hoyenga, K. T. & Aeschleman, S. (1969). Social facilitation of eating in the rat. Psychonomic Science 14, 239±241. Lowe, M. R. & Fisher, E. B. (1983). Emotional reactivity, emotional eating, and obesity: a naturalistic study. Journal of Behavioral Medicine 6, 135±149. Markus, C. R., Panhuysen, G., Tuiten, A., Koppeschaar, H., Fekkes, D. & Peters, M. L. (1998). Does Carbohydraterich, protein-poor food prevent a deterioration of mood and cognitive performance of stress-prone subjects when subjected to a stressful task? Appetite 32, 49±65. Metropolitan Life Insurance Company (1984). Height and weight tables. Statistical Bulletin Metropolitan Insurance Company 64, 2±9. Reid, M. & Hammersley, R. (1995). Effects of carbohydrate intake on subsequent food intake and mood states. Physiology and Behavior 58, 421±427. Schlundt, D. G. (1995). Assessment of Specific Eating Behaviors and Eating Style. In D. Allision (Ed.), Handbook of assessment methods for eating behaviors and weightrelated problems: measures, theory and research. Pp. 241±302. Thousands Oaks, London & New Delhi: Sage Publications. Schlundt, D. G. & Johnson, W. G. (1990). Eating disorders: assessment and treatment. Boston, MA: Allyn & Bacon. Schlundt, D. G. & Zimering, R. T. (1988). The dieter's inventory of eating temptations: a measure of weight control competence. Addictive Behaviors 13, 151±164. Schlundt, D. G., Sbrocco, T. & Bell, C. (1988). Identification of high-risk situations in a behavioral weight loss program: application of the relapse prevention model. International Journal of Obesity 13, 223±234. Schlundt, D. G., Taylor, D., Hill, J. O., Sbrocco, T., PopeCordle, J., Kasser, T. & Arnold, D. (1991). A behavioral taxonomy of obese female participants in a weight-loss program. American Journal of Clinical Nutrition 53, 1151±1158. Schlundt, D. G., Hill, J. O., Pope-Cordle, J. & Sharp, T. (1992). The role of breakfast in the treatment of obesity: a randomized clinical trial. American Journal of Clinical Nutrition 55, 645±651. Schlundt, D. G., Virts, K. L., Sbrocco, T. & Pope-Cordle, J. (1993). A sequential behavioral analysis of craving sweets in obese women. Addictive Behavior 18, 67±80. Schmitz, B. A. (1996). The relationship between affect and binge eating. Dissertation Abstracts International: Section B: The Sciences and Engineering 56, 7055. Steptoe, A., Lipsey, Z. & Wardle, J. (1998). Stress, hassles and variations in alcohol consumption, food choice and physical exercise: a diary study. British Journal of Health Psychology 3, 51±63. Tolman, C. W. & Wilson, G. F. (1965). Social feeding in domestic chicks. Animal Behavior 13, 134±142. Watson, D. & Tellegen, A. (1985). Toward a consensual structure of mood. Psychological Bulletin 98, 219±235. Zajonc, R. B. (1965). Social facilitation. Science 149, 269±274.