Many eating studies in psychology, consumer behavior and marketing journals are dismissed, because they focus on how much one serves and not how much is eaten. We develop a means of estimating the percentage of self-served food that is consumed under various conditions. An aggregate analysis was conducted of studies where participants served themselves food and where actual intake was measured. Analyses explored what percentage of food was consumed depending on population, food and situational cues and generally showed that adults consistently consume the vast majority (91.7%) of what they serve themselves. This was higher for meals (92.8%) than for snacks (76.1%) and higher when a person was not distracted (97.1%) than when he or she was distracted (88.8%). The percentage eaten did not vary between lab (90.7%) and field settings (91.9%). Because many eating behavior studies outside of nutrition measure food selection, but not intake, the aggregate estimates presented in this research can enable obesity, nutrition and public health researchers to extrapolate how much may have been eaten in such studies. Doing so will extend their relevance to better understanding eating behavior and better developing solutions to overeating.
Concerns about an increasingly obese population have prompted investigations into how increasing serving sizes1 and consumption norms may lead to increasing weight gain over time.2,3 As such, there are broad literatures on portion-size norms—largely in psychology and consumer behavior—and there are broad literatures on portion-size and actual consumption—largely in nutrition and public health.2, 3, 4, 5, 6 Unfortunately, these two literatures are largely mutually exclusive. In many portion-size norm studies, participants are asked to demonstrate their typical serving size, but they either do not actually consume the food they have served or the amount they serve is not measured. In contrast, many nutrition studies do not allow people to serve themselves but instead pre-serve them fixed, intentionally large portions. Few studies, however, have examined the percentage of food that people consume when they serve themselves. Knowing what percentage of self-served food is typically eaten under various conditions has immediate implications for helping extend key findings in psychology and consumer behavior (where intake is seldom measured) into the discussions and solutions of obesity that are taking place in public health.
Beyond providing a more realistic view of actual consumption in circumstances in which individuals serve themselves (such as when eating in the home or at a buffet-style restaurant), being able to roughly estimate of the proportion of self-served food consumed under different conditions brings new relevance to an entire generation of literature that has been previously dismissed by the nutrition and medical communities. Many researchers outside the nutrition and medical fields conduct studies related to food psychology that do not measure consumption, because the variable of interest is the amount of food selected and not how much they eat. Far from being irrelevant, these studies can add richness to what motivates and drives consumption behavior. If there were documented estimates of what percentage of food was consumed in these studies, their findings would increase in relevance and value. This research provides a preliminary aggregate analysis of the percentage of self-served food that is eaten depending on the people, the food and the eating situation.
Materials and Methods
Meta-analysis is a technique used to combine findings from different studies that collected similar measures,7 and it is a proven technique to quantify an individual’s health-related behavior within the context of the complex environment of consumption behavior.8 Crombie and Davies emphasize that a meta-analysis is only as good as the systematic literature review used to define the scope.6 Because there are too few studies to date to perform a comprehensive meta-analysis, aggregate estimates from existing studies will be computed to provide preliminary estimates by using a similar selection process as was emphasized by the 2009 PRISMA checklist for meta-analysis.9 The eligibility for inclusion of the study included the following:
English-language study of American or Canadian participants. Given the hypothesis that some aspects of consumption differ across cultures, this study was limited to an American or Canadian context.
Participants older than age three. As children younger than three do not normally have complete control of selection of foods and quantity consumed, this study did not include them (no studies were rejected based on this criteria).
Amount of food selected and consumed by participants must have been measured in grams or in a unit of measurement convertible to grams (for instance, calories of plain M&M’s). These quantities must also have been cited in the published paper or have been made available by the author.
Studies published in all years searchable in the electronic databases utilized. As this is a relatively recent area of research, the studies found were published from January 1997 to June 2013; widely cited unpublished working papers were also considered to expand generalizability.
Given the subject matter, PsychInfo, the Social Science Citation Index (SSCI) and PubMed databases were selected as the primary sources of information. Search terms included consumption volume, self-serving food portion, plate waste, clean your plate, self-served meal, family style meal consumption, food serving, food intake, food serving consumption and eating behavior. Topical search terms were utilized first; then specific author names identified by a review of pertinent literature were utilized to broaden the scope of the review. In addition, active scholars in this area were also contacted to obtain unpublished papers that could be included in these analyses. The published and unpublished working papers included in the analysis include the relevant studies in the last twenty references in this paper.10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28
Because s.ds. were not available in all of the studies, a conservative s.d. was conjectured based on the data that was provided. As a number of studies showed that in some cases as much as 95% of the food that was self-served was consumed, a 10% s.d was selected as one that would be both conservative and reasonable. Using the means, sample sizes, conjectured s.d., the t-tests and associated P-values were calculated.
Out of the 127 studies that were reviewed, our systematic review identified 14 studies (involving 1179 different subjects) meeting the eligibility criteria. Many seemingly relevant studies were rejected for not meeting the third eligibility criteria (for example, not collecting or providing consumption data or serving fixed portions to participants). These studies in this analysis involved the serving and consumption of a single food. Within these 14 studies, there were 49 distinct subpopulations differentiated by demographic factors such as gender, age and body mass index; food-related factors such as healthfulness of the foods and type of eating occasion; and environmental factors such as plate size and lab or realistic setting. Forty-three of these subpopulations were adults, classified as individuals aged >18 years (n=853), and six subpopulations were children aged 4–17 years (n=326). Of the 43 adult subpopulations, 13 ate healthy foods (n=148), 11 unhealthy foods (n=399) and 19 ambiguously healthy foods (n=306).
Overall, individuals consumed 90.69% of the food they served themselves. Importantly, while this statistic varied slightly across the demographic groups, types of foods and environmental differences, it remained remarkably high across all groups. Adults generally eat most of what they serve themselves, regardless of the type of person they are, the type of food they are eating or the environment they are in. When studies are weighted by their sample sizes, the mean percentage consumed is 83.7%. When looking at demographic differences (Table 1), male and female adults ate similar percentages of what they served (90.3% vs 91.8%; t429=1.69, P<0.09). For a large variety of reasons—familiarity, motor control, calibration and experience—it is not surprising that these adults, in general, consumed much higher percentage of what they self-served than did children (91.6% vs 59.1%; t1177=48.6, P<0.001).
The percentage of self-served food that a person ate was also related to the type of food that was served (Table 2). Consistent with USDA Dietary Guidelines, foods were coded as healthy if they were fruits, vegetables, whole grains and lean meat or dairy. Foods were coded as less healthy if they were sweet or salty processed food or higher fat meat (such as BBQ ribs or fried chicken wings) or dairy (ice cream). Using this coding system, it was found that people ate a much higher percentage of the healthy foods they self-served than the unhealthy foods (91.2% vs 80.6%; t545=14.17, P<0.001). Similarly, when comparing continuous foods (such as pasta, applesauce and ice cream) vs unitary or discrete foods (such as carrot sticks, chicken wings and cookies), people consumed a much higher percentage of continuous foods than unitary or discrete foods (93.0% vs 71.8%; txx=2.00, P<0.001). Finally, a much higher percentage of meals were consumed than snacks (93.8% vs 76.1%; t852=32.57, P<0.001).
Environmental differences also influenced the degree to which one cleaned their plate. Distracting environments could influence both how much a person serves and what percentage of it they eat. Indeed, people who were in a distracting environment—either with other people or while watching television—ate a significantly lower percentage of what they served than those who were not distracted and eating alone (88.8% vs 97.2%; t851=16.99, P<0.001). Furthermore, in studies that focused on differences in dinnerware size, those eating from small plates or bowls understandably ate a higher percentage of what they served than those eating off of larger plates or bowls (96.2% vs 93.0%; t353=3.66, P<0.001). It is important to note that both distraction and dinnerware size can influence both how much is served as well as the percentage that is eaten. Perhaps most interesting, however, is the finding that there is not a strong difference between the percentage of how much one consumes in the field vs in the lab (90.7% vs 91.9%; t851=1.9, P=0.059).
This analysis shows that individuals consistently consume the vast majority of what they serve themselves, thus strategies to decrease self-selected serving size deserve further exploration. Yet what it usefully also provides is a means to estimate how much food served in a wide range of behavioral studies was likely have been consumed based on the people involved, the food involved and the situation. For instance, suppose a psychology study showed that skipping a meal resulted in a person serving 200 more calories of carbohydrates during their next meal than they otherwise would. Based on the aggregate results presented here, we would be able to estimate how many of these 200 calories are eaten based on the person (adults eat 91.7%), on the food (93.0% of served food that is continuous, such as pasta, is eaten), on the occasion (97.1% of self-served dinner foods are eaten) and on the environment (in a non-distracting environment, 97.2% of the food served is consumed). Across all of these different individual estimates (91.7, 93.0, 97.1 and 97.2%), we could confidently estimate that at least 90% of what was served—180 of the 200 extra calories—was likely to have been consumed.
The studies used in this aggregate analysis are included in the references. Because of the scarcity of studies that took measures of both how much one served and how much they ate, this research—although providing seemingly robust findings—should be viewed as a preliminary investigation that paves the way for a more detailed meta-analysis as more studies become available for aggregation. In addition, this aggregate analysis is also limited, because it examines food intake in grams rather than in calories (which were not always reported in the studies). Furthermore, space does not permit all of the individual studies to be fully discussed and referenced in terms of contextual factors, such why there might be differences between the percentage of eaten of meals vs snacks or differences between distracting and non-distracting environments.
To the extent it would be useful to even more carefully catalogue estimates of how much served food was eaten, further studies can examine key subpopulations or situations. The best example of this would be to more closely investigate what percentage of self-served food is eaten by younger vs older children. For instance, the ages of the children in these studies were quite young, generally 4—8 years, and it is very likely that as they age beyond this level their experience, preferences and the gauging of their hunger and satiety become more precise and their intake percentage would rise. As a result, as children age, the percentage of what they eat probably rises to that of what an adult would eat.
These results provide initial evidence that key findings from previously conducted eating studies in psychology and consumer behavior should be better integrated with our current understanding even if they did not initially measure actual intake. The safest studies to do these with are studies involving adults (not children) who serve themselves meals (and not snacks). In these studies, the typical adult ate over 90% of what they served themselves.
There has been past criticism that lab studies do not reflect the reality of field studies. Importantly, these results show over 90% of what is served gets consumed in both these situations. This provides helpful justification for another dimension on which lab study results are more applicable to everyday situations than field study proponents claim.
United States Department of Health and Human Services, Public Health Service, Office of the Surgeon General. The Surgeon General’s Call to Action to Prevent and Decrease Overweight and Obesity. Office of Disease Prevention and Health Promotion; Centers for Disease Control and Prevention, National Institutes of Health: Washington, DC, USA, 2001.
Rolls BJ, Roe LS, Kral TVE, Meengs JS, Wall DE . Increasing the portion size of a packaged snack increases energy intake in men and women. Appetite 2004; 42: 63–69.
Rolls BJ, Engell D, Birch LL . Serving portion size influences 5-year-old but not 3-year-old children's food intakes. J Am Diet Assoc 2000; 100: 232–234.
Rolls BJ, Morris EL, Roe LS . Portion size of food affects energy intake in normal-weight and overweight men and women. Am J Clin Nutr 2002; 76: 1207–1213.
Wansink B, van Ittersum K . Portion size me: plate size can decrease serving size, intake, and food waste. J Exp Psychol Appl 2014; 19: 320–332.
Geier AB, Rozin P, Doros G . Unit bias. Psychol Sci 2006; 17: 521–525.
Crombie IK, Davies HTO . What is a meta-analysis? Hayward Medical Communications. Available from www.whatisseries.co.uk 2009.
Farley JU, Lehmann DR, Sawyer A . Empirical marketing generalization using meta-analysis. Mark Sci 1995; 14: G36–G46.
Moher D, Liberati A, Tetzlaff J, Altman DG The PRISMA Group. Preferred Reporting Items for Systematic Reviews and Meta-Analyses: the PRISMA statement. PLoS Med 2009; 6: e1000097.
Wansink B, van Ittersum K . Extraverted children are more biased by bowl sizes than introverts. PLoS One 2013; 8: e78224.
Wansink B, Shimizu M, Brumberg A . How vegetables make the meal: their hedonic and heroic impact on perceptions of the meal and of the preparer. Public Health Nutr 2013; 16: 1988–1994.
Van Ittersum K, Wansink B . Plate size and color suggestibility: the Delboeuf illusion’s bias on serving and eating behavior. J Consum Res 2012; 39: 215–228.
Wansink B, Tal A, Shimizu M . First foods most: after 18-hour fast, people drawn to starches first and vegetables last. Arch Int Med 2012; 172: 961–963.
Van Kleef E, Shimizu M, Wansink B . Serving bowl selection biases the amount of food served. J Nutr Educ Behav 2012; 44: 66–70.
Shimizu M, Wansink B . Watching food-related television increases caloric intake in restrained eaters. Appetite 2011; 57: 661–664.
Van Kleef E, Shimizu M, Wansink B . Food compensation: do exercise ads change food intake? Int J Behav Nutr Phys Act 2011; 8: 6.
Just DR, Wansink B . The flat-rate pricing paradox: conflicting effects of 'all-you-can-eat' buffet pricing. Rev Econ Stat 2011; 93: 193–200.
Shimizu M, Payne CR, Wansink B . When snacks become meals: how hunger and environmental cues bias food intake. Int J Behav Nutr Phys Act 2010; 7: 63.
Garg N, Wansink B, Inman JJ . The influence of incidental affect on consumers’ food intake. J Mark 2007; 71: 194–206.
Adams MA, Pelletier RL, Zive MM, Sallis JF . Salad bars and fruit and vegetable consumption in elementary schools: a plate waste study. J Am Diet Assoc 2005; 105: 1789–1792.
Branen L, Fletcher J, Myers L . Effects of pre-portioned and family-style food service on preschool children's food intake and waste at snacktime. J Res Child Educ 1997; 12: 88–95.
Just DR, Turvey CG, Wansink B . Biosecurity terrorism, food safety, and food consumption behavior: using experimental psychology to analyze economic behavior. J Agr Resour Econ 2009; 34: 91–108.
Koh J, Pliner P . The effects of degree of acquaintance, plate size, and sharing on food intake. Appetite 2009; 52: 595–602.
Wansink B, Cheney M . Super bowls: serving bowl size and food consumption. JAMA 2005; 293: 1727–1728.
Wansink B, van Ittersum K, Payne CR . Larger bowl size increases the amount of cereal children request, consume, and waste. J Pediatr 2014; 164: 323–326.
Rollins K . Working paper Kitchen Geography and Food Serving Behavior. Cornell University: Ithaca, NY, USA.
Wansink B, Chandon P . Can low-fat nutrition labels lead to obesity? Obesity 2006; 14: A49–A50.
McFerran B, Dahl DW, Fitzsimons GJ, Morales AC . I’ll have what she's having: effects of body type on the food choices of others (Study 1). J Consum Res 2009; 36: 915–929.
BW conceived of and designed the study. BW and KJ determined eligibility for inclusion in analysis. Based on these criteria, KJ contacted authors, selected the studies, compiled and analyzed the reported data. BW drafted the paper, and BW and KJ reviewed the manuscript and approved it for submission.
The authors declare no conflict of interest.
About this article
Cite this article
Wansink, B., Johnson, K. The clean plate club: about 92% of self-served food is eaten. Int J Obes 39, 371–374 (2015). https://doi.org/10.1038/ijo.2014.104
Journal of Cleaner Production (2021)
To waste or not to waste: Exploring motivational factors of Generation Z hospitality employees towards food wastage in the hospitality industry
International Journal of Hospitality Management (2019)
Frontiers in Nutrition (2018)