Calorie and Gram Differences between Meals at Fast Food and Table Service Restaurants

One of the largest changes in the American eating habits in recent decades has been the increasing reliance on food eaten away from home (FAFH). According to the United States Department of Agriculture (USDA), FAFH increased from 33% of total food expenditures in 1970 to 47% by 2003, with most occurring at table service and fast food restaurants. This growth has generated concern about the possible effect of FAFH on dietary quality. There is considerable evidence that meals eaten in restaurants are generally of lower nutritional quality than meals eaten at home, mainly due to higher fat and calorie content (Lin, Guthrie, and Frazao). The main concern is obesity, the growth of which has paralleled rising FAFH consumption. From 1971 to 2002, the prevalence of obesity among U.S. adults rose from 15 to 31%, and for children, from 4 to 16% (National Center for Health Statistics). Many studies relating obesity and energy intake with FAFH have been conducted (Haines et al.; McCrory et al.; Binkley, Eales, and Jekanowski; Ma et al.).

A feature of the debate on the nutritional consequences of FAFH is that the emphasis is on fast food. Examples in the popular media include Eric Schlosser's best-selling book Fast Food Nation, which was made into a movie, and the successful documentary Supersize Me. Fast food firms figure prominently in Fat Land: How Americans Became the Fattest People in the World, Critser's indictment of the U.S. food industry. Many of the studies by nutritionists dealing with obesity are confined to fast food. This is especially true in the case of studies on children (Ebbeling et al.; French et al.). A recent paper on childhood obesity in a leading British medical journal mentions fast food 16 times, including specific recommendations to tax fast foods and require them to carry nutritional labels. Full service restaurants are not mentioned (Ebberling, Pawlak, and Ludwig). Other studies have concluded that fast food firms place restaurants near public elementary schools and in minority neighborhoods, presumably resulting in poor eating patterns (Austin et al.; Block, Scribner, and DeSalvo). A Canadian study found an association between fast food restaurant intensity and cardiovascular disease in Ontario (Alter and Eny). Studies such as these are often featured in news reports, furthering concerns focused on fast food. It is not surprising that a 2004 poll found that 43% of respondents thought that fast food bears a “great deal” of responsibility for the obesity crisis, and that it was not felt necessary even to ask about other forms of FAFH (W.K. Kellogg Foundation). Recent studies have found that more nutrition-orientated consumers tend to avoid fast food, but show no reluctance to patronize table service restaurants (Binkley; Stewart et al.).

Although there is a large body of evidence that fast food is nutritionally inferior to meals prepared at home, there is little, if any, direct evidence that the nutritional consequences of fast food substantially differ from those of the other forms of FAFH. No study of which we are aware has made the comparison. It is the purpose of this study to do so. Using a large household dietary survey, we examine the per meal and per day intakes of food by weight and calories. We then compare how these vary by where meals are eaten. We consider only fast food, table service, and home meals and focus on a comparison of fast food with table service restaurants, and these with meals at home. We find that consumers tend to eat less and tend to obtain fewer calories when at a fast food restaurant than at a table service restaurant, and that both exceed meals at home. However, we find that consumers are more likely to compensate for large table service meals by eating less during the rest of the day than what is true of fast food. Thus, in many cases, fast food meals do result in larger differences for the day as a whole, but more for adults than for children. This is likely because of their greater energy density.

Data and Methods

The data used are from the USDA 1994–96 Continuing Survey of Food Intakes by Individuals (CSFII) (Tippett and Cypel). At the time of this study it was the most recently available large sample of nutritional intake data with the level of detail needed for this study and has been used in numerous studies on the nature and consequences of the American diet (Lin, Guthrie, and Frazao; Binkley, Eales, and Jekanowski; Wilde, McNamara, and Ranney). The CSFII involves 16,103 individuals, most of whom supplied two nonconsecutive days of data, collected using 24-hour recalls conducted by trained interviewers. The nutrient intake lists the name of each food eaten, a detailed breakdown of its nutritional content, where it was obtained, and whether it was eaten at breakfast, brunch, lunch, dinner, or as a snack. The data also include demographic measures for the individuals surveyed.

The food consumption data consist of slightly fewer than 450,000 observations, each representing a food item eaten on a survey day. These account for about 135,000 pounds of food and drink, and nearly 58 million calories. Table 1 shows the proportion of these obtained at home, at table service and fast food restaurants, and from other sources (e.g., schools, cafeterias, bars, vending machines, etc.).¹ As can be seen, table service and fast food together account for 13% of food eaten but more than 16% of calories. The difference is especially large for fast food, showing it to be the most energy dense of the three main types.

We confined the analysis to breakfasts, lunches, and dinners and required the meal to have at least two items, all from the same place. Thus, for example, a meal with take-out pizza and a homemade salad would not be included. This was done so that meals could be unambiguously classified as to the source and the time of day. For the full-day analysis, only days with a single FAFH meal or days with no FAFH were included, thus yielding an estimate of the effect of a single restaurant meal on a day's consumption of grams and calories.²

The analysis was performed, by meal, separately for adults, teenagers (age 13–18 years), and children (12 years and under). The distribution of meals is shown in table 2. Not surprisingly, children and teenagers make greater use of fast food than table service, while adults do the opposite. The percentages eaten in both types of restaurant are 8.2, 18.0, and 16.6 for children, teenagers, and adults, respectively.³

The bulk of the analysis was conducted using ordinary least squares regression, accounting for the CSFII survey design with data set weights using the SurveyReg procedure in SAS (SAS Institute, Inc., Cary, NC).⁴ We estimated equations explaining variation in the intake of calories and total food weight in grams at the meal as a function of several variables, focusing on the indicator variables table service and fast food. Home meals were the omitted reference group, so the coefficients on the indicators are direct measures of the differences with home meals. The remaining variables are of little direct interest and were included to control the error and reduce the possibility of omitted variable bias. These variables are female, an indicator variable taking the value 1 for female respondents; age in years; income in thousands of dollars per household member; education, measured as the number of years of schooling (for teenagers and children, this measure referred to their parent(s)); and the ethnic indicators African American and Hispanic. Because on weekends people are likely to eat somewhat differently than they do during the week, three indicator variables Friday, Saturday, and Sunday, taking the value 1 if the meal occurred on the respective day, were included in the equation. Finally, all respondents were asked whether their food intake on the survey day(s) was more or less than normal. We included the indicator variables more and less, which took the value 1 when the corresponding answer was in the affirmative.

Results

The regression results appear in tables 3–5, for adults, teenagers, and children, respectively. To save space, the upper part of each table contains only the estimated coefficients for each meal for grams and calories, with statistical significance indicated by one, two, or three asterisks for 0.10, 0.05, and 0.01, respectively. The contrast row in the lower part contains the difference , where ts and ff denote table service and fast food, the number of asterisks again indicating the degree of statistical significance. For the full-day models in these tables, no distinction is made as to whether FAFH meals are breakfast, lunch, or dinner. This is mainly because doing so revealed no meaningful differences among the meals, and presentation of all models would lead to an overabundance of tables.

The key results, of course, are those for the binary variables table and fast food. Their estimated coefficients measure the differences between restaurant and home meals. Thus, for example, from table 1, we see that, in the sample, an adult breakfast at a table service restaurant is estimated to have 139 g, or almost 5 ounces, more food than a breakfast at home, with about 261 additional calories. The corresponding values for fast food also show increases: 43 g and 206 calories. For all three meals, the calorie increases for both table service and fast food are substantial and significant at 0.05.

The contrasts directly measure the table service-fast food differences. From these, we see that, for all three meals, adults eat significantly more food at table service than at fast food restaurants, and obtain significantly more calories in two of the three meals, the exception being lunch. However, for the entire day, although both types of FAFH are associated with more grams and calories, the differences for table service are not large, and only those for fast food are significant. The contrast shows a significantly larger (139) average increase in calories for fast food. Overall, there is little indication that adults compensate for fast food excess by making adjustments at other meals, but they evidently do so with table service meals.

From tables 2 and 3 we see that, in all cases, children and teenagers eat significantly larger meals, with significantly more calories at table service restaurants than at fast food outlets. The differences are particularly large for children. Again, however, there is evidence of greater compensation when the restaurant meal is table service. The day differences are smaller, and for teenagers, the average effect of fast food is worse than that of table service for both grams and calories, although insignificantly so. Somewhat surprisingly, children are the only group for whom the average day differences are larger for table service than for fast food. However, the differences are small. That for calories is 5, a number that can hardly be considered meaningful, despite its statistical significance. Note also that the coefficient on fast food is significant but that on table service is not, despite being larger.

Regarding other variables, there are no surprises. Age has a positive effect for children and teenagers and a negative effect for adults. Women tend to eat significantly less than men, obtaining fewer calories. Most of the remaining variables display no consistently strong impacts. One of the most interesting effects is that, although African Americans generally appear to eat more at a restaurant meal, in the sample, the impact for the entire day appears to be negative. The latter appears to apply to Hispanic diners also. Finally, on balance, the coefficients on less tend to be smaller in absolute value than are those on more. This suggests people are more sensitive to undereating than to overeating.

Because of the observational nature of the data, the possibility exists that these results reflect specification problems, such as neglected heterogeneity and especially selectivity. One example was noted above—people with nutrition concerns are more likely to choose table service than fast food restaurants. One way to add a degree of control is to observe the same subject under different conditions. To a limited extent that is possible with our data, and we did so in two ways, one involving a direct comparison of fast food and table service, and a second involving a day with and without FAFH.

For the first, there were 298 cases where, on one survey day, an individual ate a meal at a table service restaurant, and on the other survey day, ate the same meal at a fast food restaurant. Comparing these two meals controls for time-invariant differences specific to individuals. The grams and calories for these meals are shown in table 6, for the meal and the day. Because of the small number of observations, ages are aggregated (although the large majority is adult). The observations for days are especially short because many of the 299 individuals had FAFH at times other than the meal on one or both of the days. Examining the differences, we see that for this small sample, the respondents on average ate more at a table service than at a fast food restaurant for all meals, and only for lunch did they consume more calories of fast food. However, only grams at dinner are significant (although breakfast calories have a significance value of 0.106). These are similar to what is indicated by the contrasts in the regression tables. For days, the results are somewhat different from those above, especially for calories, for which the differences are larger in absolute value. On average, fast food is worse, but this is evidently all due to fast food lunches, the only significant day effect. The small sample and consequent lack of strong significance in these results, even when aggregated over all meals, precludes any firm conclusions. What can be said is that they do not greatly differ from those for the full sample above.

In the second case, we compared food intake by a given individual on a day with FAFH to a day with none. From our main sample, we selected the subsample that met the following criteria. First, respondents provided two full days of data. Second, on one of the days, each individual had exactly one restaurant meal, and no restaurant meals on the other. Third, the meal eaten at a restaurant (breakfast, lunch, or dinner) was not skipped on the day without FAFH. Finally, on neither of the days was food intake reported to be more or less than usual. We then examined each individual's calorie and gram differences across the two days for the meal in question and for the entire day. We retained the original age groups. However, because sample sizes were small, we aggregated over the three meal occasions. Table 7 shows the differences.

For the meal, the results for children and adults are similar to those in tables 3 and 5, although the magnitudes of the FAFH meal-home meal differences appear somewhat smaller. As before, the differences are greater for table service than for fast food. For teenagers, however, the story is somewhat different. Now, the differences between FAFH and home meals are larger than before for fast food, certainly for calories, and fast food exceeds table service for both calories and grams. We attribute this discrepancy to the inherent variability of this data and the relatively small size of this subsample. Note, for example, that although teenagers eat much more food at both fast food and table service meals than at home, neither difference is significant.⁵

With some exceptions, the FAFH-home day differences are smaller than the meal differences. This again suggests some compensation for FAFH meals by reduced intakes during the rest of the day. In only two cases are the day differences significant at 0.10 or better: calories for teenagers at both restaurant types. For adults and, to a lesser extent, for children, compensation appears to be greater for table service than for fast food. However, in all cases, the day differences for table service remain the larger of the two. This differs from the results in tables 3–5, where this is the case only for children.

Taken as a whole, the results in tables 3 through 6 suggest that larger meals are consumed at table service than at fast food restaurants, but less consumption at other times during the day for table service narrows, and often reverses, this difference. Insights into these outcomes are provided by information shown in table 8. Much of the concern with fast food stems from its fat/sugar and hence its calorie content. As the table indicates (and which was seen above in table 1), in the large majority of cases, the energy density of fast food is indeed higher than that of table service food, and far greater than that of home food. Previous studies have provided evidence that, given the calorie content of a meal, the lower its energy density, that is, the greater its “bulk,” the more likely an individual will feel satisfied and not desire to eat at other times during the day (Poppitt and Prentice; Prentice and Jebb). The lower compensation for a fast food meal evident in this data bears this out.

A second issue with fast food is the large sizes of items, often attributed to “supersize” pricing, and simple economics suggests this will lead to overeating. As the table shows, in every case, the average weight of individual fast food items is indeed the largest of the three sources, usually substantially so. But, as the table also shows, in all instances considered, the average fast food meal has fewer items than a meal at a table service restaurant, resulting in less total food and fewer calories. Furthermore, in every case, the average table service meal has more items than the corresponding meal at home, resulting in more calories. Thus, while fast food meals may be large due to item sizes, table service meals appear to be at least as large, due to the number of items. In either case, the result is meals with more calories than home meals.

Both forms of “supersizing” are responses to the same market forces. The cost of a meal comprises labor, overhead, and ingredients. Labor and overhead costs are nearly invariant with the size of the meal, so meal size can be increased with a less-than-commensurate increase in price. This results in a considerable pressure to increase meal sizes. In a market dominated by competitive national chains in nearly all segments, the size of the meal is an easily implemented competitive tool (DiDomenico). This should be truer for table service than for fast food because labor and overhead costs are generally higher for table service restaurants. Hence, increasing the meal size can be done at a relatively low cost. Failing to do so can result in customer dissatisfaction, and a dissatisfied customer may not return to any stores in the chain after an unsatisfactory meal at any one of them.

It is perhaps ironic that the differences in pricing methods by fast food and table service restaurants may contribute to the meal size differences in the results above. At the time of placing an order, a fast food customer has more control over the amount of food purchased, and if a supersize is desired, it must be overtly chosen. The a la carte pricing predominant in the fast food industry makes each meal item a separate budgetary decision. At most other restaurants, items tend to be “bundled” under a single dinner price. Often meals are accompanied by large side servings of low cost, high-calorie foods, such as rice and pasta. Studies have shown that consumers tend to passively eat the entire portion of servings of varying sizes of items such as pasta, and are unable to detect size differences after they have consumed them (Rolls; Diliberti et al.). In contrast to this, one can hardly claim passivity after ordering a fast food item described with terms such as “big,” “monster,” “double,” and so on. The extreme case of bundling is the “all-you-can-eat” special, which is standard practice in buffet style restaurants: once the admission price is paid, the consumer's marginal outlay is zero. The same pricing method is often used for meal components, with ad libitum availability of salad, bread, and even desserts. This kind of pricing obviously encourages overconsumption.

Concluding Remarks

Fast food is typically regarded as of lower nutrition and higher in calories than other restaurant fare. We found this to be true, in the sense that, on average, fast food in our study was more energy dense than food from a table service restaurant. However, we also found that fast food meals tend to be smaller. Consequently, the typical fast food meal in this study had fewer calories than the average meal from a table service restaurant, be the diner an adult, a teenager, or a child. But, we also found that fast food diners are less likely to reduce their food consumption during the rest of the day than are those eating at table service restaurants, most likely because of the difference in energy density. As a result, fast food may ultimately result in more calories. However, this result was not as robust as that for meals, and was not true of the subsample in which the same individuals were compared on days with FAFH and days without FAFH. Perhaps the most surprising result of the study was our finding that fast food had the largest effects for adults, and that children's caloric intakes for both meals and days was greatest when they ate at table service restaurants.⁶

Before concluding, we note two caveats associated with these outcomes. Regarding meal sizes, there is the possibility that diners may choose tables service restaurants for larger meals and fast food when they wish to eat less. Although the regression variables measuring whether food intake was normal on survey days may partially account for this, there is no obvious way to eliminate it, including our analysis using paired data. A second caution is that in the time after the sample was taken, some fast food chains have been featuring salads and other fruit/vegetable-based fare. These have met with a degree of success, and this would be expected to lower the average energy density of fast food, perhaps reducing the difference between fast food and table service due to differences in compensation.

Our main conclusion from the study is that, in an all-encompassing sense, there is not likely to be a great deal of difference between table service and fast food restaurant meals in terms of their caloric intakes. Both tend to increase calories relative to home meals, and while the net effect after compensation may not be large, and that for table service smaller than that for fast food, even a small increase in caloric intake, taken over an ever-increasing number of FAFH meals, can have a large cumulative effect. Certainly, the lesson from this study is not that fast food is good for you. However, avoiding fast food when dining out is no assurance of avoiding extra calories. The lesson is that it is misleading to focus concerns about the nutritional effects of increased FAFH primarily on fast food. All FAFH should be considered.

Figures and Tables