Bachelor thesis
Name: Harm de Jong
Student number: 5933617
Study: Economics and business
Field: Behavioral economics
Supervisor: Jindi Zheng
Coordinator: Marcel Boumans
The Sunk cost fallacy:
Does the price of a menu in an
all-you-can-eat restaurant have an effect
on the amount of food customers
eat?
Abstract
Does price have influence on the amount of food customers eat in an all-you-can-eat restaurant? Previous literature found that if the price of an all-you-can-eat menu is higher, customers eat more (Just & Wansink 2011). According to neoclassical economic theory customers should treat the all-you-can-eat menu price as a sunk cost since they cannot recover it. In this research I observe customers in an all-you-can-eat sushi and grill restaurant where the price of an all-you-can-eat experience is higher in on a weekend day. I did not found a significant difference on the amount of food eaten on the weekend day and the weekday. This paper gives an insight in when and if people treat sunk costs as relevant in their future economic decisions.
Introduction
Sunk costs are costs which are made and cannot be recovered. Neoclassical Economic theory suggests these costs may never be a factor in future economic decisions (Frank and Bernanke 2006, 10; Mankiw 2004, 297). Rationality is one of the main assumptions in neoclassical economic theory. If a cost is made an cannot be recovered this cost does not have any influence on future economic benefits. It should therefore be ignored when deciding to make an investment. Past literature suggests that these costs sometimes however are taken into account (Garland 1990; Arkes & Ayton 1999; Arkes & Blumer 1985; Tversky & Kahneman 1981). This is irrational according to neoclassical economic theory and is called the sunk cost fallacy. Thaler (1980) was the first to cite anecdotal evidence of the sunk cost fallacy. Friedman, Pommerenke, Lukose, Milam & Huberman (2005) wrote an article where they search for the sunk cost fallacy. They give a famous example of a concertgoer who paid for a ticket but realizes after 5 minutes into the show that he finds it horrible. He sticks around however because he wants to get his “money’s worth”. This concertgoer uses the sunk cost of the concert ticket to make his decision to stay and so he must endure the horrible show.
Friedman, Pommerenke, Lukose, Milam & Huberman (2005) find that it is very hard to find the sunk cost fallacy. They give examples of existing “evidence” of the sunk costs fallacy like the invasion in Iraq in 2003. The Iraqi dictator had agreed with weapon inspections and the placement of NAVO troops but the Americans had already invested tens of billions of dollars in getting ready for the invasion and disrupted the lives of thousands of soldiers so they invaded anyway. There is also another explanation possible like that the American policymakers were scared that cancellation of the attack could hurt the credibility of the united states. Friedman, Pommerenke, Lukose, Milam & Huberman (2005) suggest that it is very hard to isolate the sunk cost fallacy in these situations. An example in Holland is the noord-zuidlijn in Amsterdam. This is a metro line between north and south Amsterdam which turned out to cost a lot more than expected. After investing millions in the project there was a discussion whether to invest millions more or stop. They decided to invest millions more to finish the line. Like the invasion in Iraq the question is whether they chose to finish the line because they already invested millions of euro’s or for other reasons like the possible future benefits for transportation in Amsterdam.
Most of the evidence of the sunk cost fallacy has been given in studies were experiments were done in a laboratory (Staw 1976, 1981, Arkes and Blumer 1985, Whyte 1993, and Khan, Salter, and Sharp 2000) These studies find that if an agent already invested money in a project they tend to invest more in that particular investment.
In this research I will search for the sunk cost fallacy in all-you-can-eat restaurants. In all-you-can-eat restaurants you pay a fixed price and can eat as much as you like. The fixed price is thus a sunk cost. The price should not have influence on the amount of food a customer eats. This research is
important because obesity is becoming a large problem especially in the western world. Levitsky, Halbmeier &Mrdjenovic (2004) show that 20% of the weight gain of a college freshman is due to all-you-can-eat dining. Wang, Beydoun, Liang, Caballero & Kumanyika (2008) show that if the weight gain in the united states continues the way it does now, all the Americans will be overweight or obese in 2048. With these statistics we could already conclude that people eat more in all-you-can-eat restaurants than they should. The question is whether the fixed cost of such a restaurant plays a role in the decision to eat.
Just and Wansink (2011) did research into eating behavior in an all-you-can-eat pizza restaurant. They investigate if people eat more of the pizza buffet when the price of the buffet is higher. If this is the case this would prove that people do not treat the fixed cost as sunk. Their theory is as follows; a persons should continue eating in an all-you-can-eat restaurant until their marginal utility of
consumption is zero according to economic theory. This is the moment when your full or satisfied. When the person continues eating after that, the marginal utility of consumption becomes negative, which means that eating becomes an unpleasant experience. The customer might enjoy the food less or has an unpleasant full stomach. Why do people tend to overeat in all-you-can-eat restaurants? Thaler (2004) suggests that customers might be motivated by the desire to “get a good deal” or “get one’s money’s worth”. He calls this transaction utility. If transaction utility is a factor in maximizing total utility customers maximize transaction and consumption utility. To do this, the sum of their derivatives must equal zero. The derivative of transaction utility is positive, for if a person eats more, he or she gets a “better deal” so more transaction utility. The more the person eats, the more transaction utility, so at the maximum the derivative of transaction utility is still positive. This means the derivative of consumption utility must be negative at the maximum which shows that transaction utility could be a factor which explains why people eat more after their marginal utility of
consumption reaches zero.
Just and Wansink (2011) conducted an experiment in an all-you-can-eat pizza restaurant. They asked the customers who came in if they would want to fill in a questionnaire. Half of the customers were given a coupon for a 50% discount for the fixed price of the pizza buffet and some coupons for free drinks and the other half just got the coupons for free drinks. They found that the group that paid half price ate less pizza (2.94 slices versus 4.09;, p-value = 0.013). Their conclusion is that when price in an all you can eat buffet is higher, consumption is higher as well, probably due to transaction utility. They also found a negative correlation between consumption and the taste of the pizza. When the customers perceived the taste of the pizza as bad, they ate more pizza. This could be because you need more bad pizza to “get your money’s worth”. If the pizza tastes good you need less pizza to get satisfied in transaction utility terms.
Siniver and Yaniv (2012) built on this theory and found that customers do not always overeat beyond the point of fullness. This happens only when the behavioral “get one’s money’s worth” constraint lies above the point of fullness. If the constraint lies below the point of fullness, the customer will just stop eating when reaching the fullness point. In this last case the customer already gets his money’s worth before reaching the fullness point so he eats until full. Siniver and Yaniv(2012) tested this by doing an experiment in a sushi restaurant on campus where there is usually an ”a la card” menu with
sushi’s for NIS 2. They organized an all-you-can-eat menu for NIS 45 and surprised the customers by giving one third a price of NIS 30 and one third a price of NIS 20. They found that the group that paid the original NIS 45 ate 24,50 sushi’s and the groups of NIS 30 and NIS 20 ate 18,56 and 18,33 sushi’s respectively. This is consequent with their theory. At NIS 30 and NIS 20 they eat about the same, because the behavioral constraint lies below the point of fullness. After respectively 15 and 10 sushi’s the customers already “get one’s money’s worth” so they just eat until they are full. At the price of NIS 45 they only get their money’s worth after 22,50 sushi’s so they eat more than the point of fullness (which probably is around 18,50 sushi’s on average).
Siniver, Mealem & Yaniv (2013) wrote a paper about paying before or after you eat in an all-you-can-eat restaurant. They suggest that paying before could be perceived as disrespectful trall-you-can-eatment. They make transaction utility broader than just the amount of food one has to eat to get “one’s money’s worth”. The way they are treated by the staff and the overall experience in the restaurant also accounts for some part in the transaction utility. They conducted an experiment in an all-you-can-eat sushi restaurant where one half of the customers had to pay beforehand and the other half had to pay afterwards. The group that paid afterwards ate 4,5 units of sushi less than the group that had to pay beforehand. This suggests that customers indeed perceive the paying beforehand as
disrespectful treatment and have to eat more to reach their maximum utility.
In my research I tried to find the sunk cost fallacy in an all-you-can-eat sushi and grill restaurant in the Netherlands. All-you-can-eat sushi restaurants become increasingly popular in Holland. Most of them have different prices during weekends. In the weekends they are usually more expensive, probably because the customers are attending the restaurants in the weekends anyway. This research is based on the research in the pizza restaurant by Just and Wansink (2011). The price difference in my experiment however is known in advance. Just and Wansink used discount coupons, so the price was a surprise for the attending customers. In my research the customers know in advance whether they are going to pay the weekend or the weekday price. Also the difference in price is smaller than in Bryan and Wansink their observation. The price during the weekends is €22,80 and during weekdays it’s €19,80. I will try to find a relation between price and consumption even when the price difference is small. While I expected to find that customers eat more during the weekend than during weekdays because of the higher price, I found that customers ate less on a weekend day. On Wednesday the customers ate an average of 15,28 dishes per person per table and on Friday the customers ate an average of 13,85 dishes per person per table. The difference of 1,43 dishes was not significant. So when the price was higher, the customers ate less (not significant) Data collecting
I collected the data during two days, one weekday and one weekend day, (Wednesday and Friday) in an all-you-can-eat sushi and grill restaurant in almere in the Netherlands called Sapporo. The all you can eat sushi and grill menu is the only menu you can order here. There is no a la card menu so everyone who is going to this restaurant knows they are going to have an all you can eat experience. On weekdays the price of an all you can eat experience is €19,80, on weekends the price is €22,80. Except for the price everything is the same on both days; the menu does not change. Customers come in and order dishes by filling in an ordering card. There is one such card per table. The dishes are numbered and customers can see pictures of the dishes on a side card. On the ordering card they can fill in the amount of dishes they want. There are 6 rounds to order and you can order a maximum
of 5 dishes per person per round. You can only order a new round when you have finished the round before. Waste is not an option. If you order something you don’t eat you have to pay a fine for that dish. On my two evenings there has not been such a fine so we know that there was no waste and they ate everything they ordered. I asked the management for permission to do the research. They would let me collect all the ordering cards but they would not let me do a questionnaire at the end of everyone’s dinner. In exchange for taking some of their time I opted to wait tables on both evenings. I arrived both days at 4pm and the restaurant opened at 5pm. The weather on both days was warm (25 degrees) and partly sunny, partly cloudy. I was wearing professional Sapporo waiting clothes so the customers did not now there were being observed. All Sapporo personnel did know I was there for research. When the customers came in I observed them and wrote down their characteristics (age, gender, thickness) and their table number. Since I was not allowed to do a questionnaire I estimated age and thickness. I made age groups so I could quite accurately decide in what age group the persons would be. I created a scale from 1 to 4 for thickness where 1 is skinny, 2 is normal sized, 3 is a somewhat fuller figure and 4 is obese. After the research I decided to drop this characteristic thickness since this is a highly problematic one. The numbers are solely based on my judgment and therefore subjective and unreliable. Studies have shown that there is a large correlation between age and weight (Kuczmarski, Kuczmarski & Najjar, 2001) so the characteristic age is enough to control for size biases. I also found a strong correlation (0.508 significant at a 0.01 level) in my own
(controversial) thickness and age observations (Figure 1 in appendix). During the evening I waited tables and made sure I wrote down all the characteristics of the people on a particular table. I instructed the personnel to make sure all the ordering cards were given to me when the table was finished. All the ordering cards were accounted for. There were 33 tables on Wednesday and 28 tables on Friday. Usually Friday is more busy than Wednesday. The personnel also told me that this particular Friday was quiet. This could be because of the warm weather (for Dutch experience). The weather was the same on Wednesday however. On Friday there was one table where I did not write down all characteristics and one table where I did not write down the age. On Wednesday all the characteristics are accounted for. There was one table on Wednesday that ordered a few dishes directly without writing them down on the ordering card. I wrote down these orderings and accounted for them by adding them to their dishes on the ordering card.
One problem I faced is the possibility of a self selection bias. Because I did research on two different days with a different sample there is a possibility that the groups of people are different as well. It could be for example that people that eat in sushi restaurants during the week are different people that eat in sushi restaurants during weekends. It could also be that a different population eats during weekdays because the price is lower. The most ideal situation would be that this research is done with the same population on one evening were half of the customers pay a different price. This is how Just and Wansink did their research in the pizza restaurant. I did not have the possibility to do my research this way because the restaurant would not let me charge different prices on one evening. This would cost the restaurant money and I do not have any money to compensate the restaurant. To control for this bias I observed the characteristics of the customers so I can account for differences in populations as much as possible.
Results
Since I only know how many dishes are eaten per table and not per person I computed average variables per table. The variables are average number of dishes per table, average age per table and the average percentage of males per table. The average amount of dishes per table is simply the total amount of dishes ordered per table divided by the number of persons at a table. I estimated the age of the customers by categories such as 11-18 years old, 18-30 years old and so on. I calculated the average age per table by taking the averages of the categories and divide the sum of that by the number of persons at the table. The percentage of male’s per table is the amount of male’s per table divided by the total number of persons at a table. The scatterplot of the amount of dishes on
Wednesday and Friday is shown in figure 2 in the appendix. The descriptive statistics are below:
Descriptive Statistics
wed or fri N Minimum Maximum Mean
Std. Deviation WED average number of dishes 33 6,00 29,50 15,2752 4,20789 average age 33 15 66 31,24 10,259 %male 33 0,00 100,00 37,3939 32,31286 Valid N (listwise) 33 FRI average number of dishes 28 6,33 20,50 13,8461 3,30606 average age 26 24 58 37,65 8,957 %male 27 0,00 67,00 31,5926 23,20189 Valid N (listwise) 26
On Wednesday the customers ate 15,28 dishes per person per table on average. On Friday they ate 13,85 dishes on average per person. So the customers ate more on average on Wednesday when the price is less than on Friday. This was not expected. I computed a t test for the difference in means:
Independent Samples Test
t-test for Equality of Means
t df Sig. (2-tailed) Mean Difference Std. Error Difference 95% Confidence Interval of the Difference Lower Upper average age Equal variances not assumed -2,559 56,353 ,013 -6,411 2,505 -11,429 -1,394 %male Equal variances not assumed ,808 57,120 ,423 5,80135 7,18179 -8,57928 20,18198 average number of dishes Equal variances not assumed 1,484 58,685 ,143 1,42908 ,96276 -,49762 3,35578
The difference between the average amount of dishes on Wednesday and Friday of 1,43 dishes is not significant. The difference in gender on Wednesday and Friday is also not significant. The difference in age is significant at a 0.05 level. The customers were older on average on Friday. The correlation between age and the average amount of dishes eaten is small and insignificant however (figure 3 in appendix). Regression analysis with average number of dishes as the independent variable,
Wednesday and Friday as the dummy dependent variable and age and gender as control variables gives the following results:
Coefficientsa Model Unstandardized Coefficients Standardized Coefficients t Sig. B Std. Error Beta 1 (Constant) 16,208 2,778 5,835 ,000 average age ,019 ,053 ,049 ,359 ,721 %male ,029 ,018 ,218 1,661 ,102 wed or fri -,656 ,533 -,169 -1,230 ,224
a. Dependent Variable: average number of dishes
The effect of Wednesday or Friday on the average amount of dishes is insignificant. This result is the same as the t test for the difference in means. We must conclude that the difference in average amount of dishes on Wednesday and Friday could be zero for the whole population, so people do not eat more dishes on weekends when price is higher. Age has no effect on the amount of dishes eaten. Gender has a small but insignificant effect on the average amount of dishes eaten. We have seen however that the difference in gender between Wednesday and Friday was insignificant.
There is a difference in the dishes that can be ordered. One can order different kind of sushi’s but also grill items like beef rolls, tenderloin and chicken skewers and fried items like spring rolls, fried duck and tempura shrimps. The difference in dishes could be a bias in the results. If a table only orders grill items they are likely to order less dishes than if a table would only order sushi dishes since sushi’s are lighter and have less calories then grill items. To account for this I have categorized the dishes into 4 categories. Suhsi’s, grill items, fried items and miscellaneous. The last category consists of fried rice, soups and salads which aren’t ordered much. The percentage of sushi dishes per table does have influence on the average amount of dishes ordered as well as the percentage of fried dishes (Figure 5). The more sushi is ordered the more dishes on average per table are ordered and the more fried dishes are ordered the less dishes on average per table are ordered. This is not a strange outcome since sushi’s are a relatively light dish and fried dishes are rich and fat. The
percentages of grill and miscellaneous dishes do not have a significant influence on the average amount of dishes ordered per table. I computed t-tests for the differences in means of percentages of the categories of dishes on Wednesdays and Fridays which are shown in figure 6 in the appendix. As we can see there is no significant difference in the percentages of categories of dishes ordered on Wednesday and Friday so we can assume that the average partition of different kind of dishes is the same on Wednesday and Friday. This means the difference in dishes does not bias the results. In figure 7 in the appendix a regression is shown where the percentages of categories of dishes are added. The result for difference between Wednesday and Friday is still insignificant.
Discussion
Because the insignificant result of the difference in average dishes on Wednesday and Friday we might conclude that price does not have influence on the amount of food eaten in an all you can eat restaurant. The customers even ate less dishes on average on Friday (not significant). Just and Wansink (2011) did find a significant result in their all you can eat pizza research. They did find that price had influence on the amount of food eaten. Explanation of the difference might be that Just and Wansink used a discount coupon when the customers came in. The price difference was a shock for the customers, they expected to pay full price and received a coupon after they made the decision to pay full price in the pizza restaurant. In my research the price difference is known in advance. Arkes and Blumer (1985) found that time has effect on the sunk cost fallacy. They
conducted an experiment where they assigned random prices to season theater tickets. They found that the ones that paid a higher price attended more shows in the first half of season, but about the same in the second half of season. Thus knowing the price difference in advance might have an effect on the sunk cost effect. Also the price difference in the pizza restaurant research is larger. The discount coupon was 50% of the price while the price on Wednesday in the sushi restaurant is only 13% less than on Friday. The difference in price in the sushi restaurant might be too small to have an effect on the eating behavior of the customers.
The difference in results between my research and the research of Just and Wansink (2011) might also be because of the self-selection bias. There is a high propability that my results are insignificant because the customers are different in nature. Despite the characteristics I observed customers that attend the restaurant in the weekends might be different in nature from customers that attend during weekdays. It could be for example that customers that eat during weekdays attend because of the lower price. These people might be so price oriented that they still eat a lot to get their money’s
worth so they end up eating more than customers during weekenddays. This could be a reason that I found (however insignificant) that customers eat more on a weekday than on a weekendday.
If people would take the price of the all you can eat menu into consideration when deciding to order more dishes we would have seen a significant difference in the average amount of dishes ordered on Wednesday and Friday. It could also be the case that the behavioral constraint was below the point of fullness like Siniver and Yaniv (2012) suggested. The price of this sushi restaurant might be so low on both days that customers have the feeling that they “get their money’s worth” before their full, so they just eat until the point of fullness. If this is the case they might have eaten more if the price was above the point where customers have the feeling that they got “their money’s worth”. This could be further researched in a similar sushi restaurant by investigating whether customers eat more when the price is even higher than on weekend days, maybe on a public holiday.
There could also be another explanation for the results. When I did the research I noticed that there were a lot more “regulars” on Wednesday. This could be evidence of a self selection bias. I do not have any data on how often the customers attended this or another sushi restaurant but I noticed on the days I was there that the staff greeted more people on Wednesday than on Friday. People that do not go out for dinner that much usually go in the weekends when they go. It could be the case that the customers on the weekend day had less experience with all-you-can-eat sushi restaurants. I suggest that if you are a regular at an all-you-can-eat restaurant you like to eat a lot. If my
observation is right it could be that there were more regulars on Wednesday who, because they are regulars and like to eat a lot, eat more than non-regulars. This would explain why the average amount of dishes ordered on Wednesday is higher than on Friday. The behavior of regulars compared to non-regulars could also be subject to further investigation.
Siniver and Yaniv (2012) also suggest two additional motivations of how much to eat in an all-you-can-eat restaurant. The first one is the all-you-all-you-can-eat teaser which is a psychological feeling a customer could have. This first motivation is described by a blogger as follows:
‘Ok eater,’ says the restaurant, ‘I dare you. Come here, and try to eat all you can eat. I bet you won't finish more than two plates.’ (Roberts, 2004).
The second motivation is to “beat the system”. This is described by another blogger as follows: Basically, my goal from the moment I walk into the buffet is to eat so much
food that the restaurant loses money. I want to eat so much that when they see me come back the next time, they get scared. I want them to worry that if I eat at their buffet too often, they might have to close it down." (Brooks, 2007). Conclusion
In this paper I searched for the sunk cost fallacy in an all-you-can-eat sushi and grill restaurant. Where previous literature found that price did have effect on the amount of food people eat, I did not find a significant result. I found that people ate less on the weekend day when price was higher (not significant). There are four reason why this result could be different from previous literature. The first is that the price was known in advance. This could have influence on the eating behavior of the customers because time has effect on the sunk cost fallacy. The second is that the price
difference was relatively small. It could be that the difference was too small to have a significant effect on the amount of food eaten. The third is that the price was too low on both days. This means the customers already felt they got their money’s worth before they were full, so they just ate until
they were full. The fourth is that there was a self-selection bias. The hunch I had was that there were more regulars on the weekday (not proven). If we assume regular all-you-can-eat restaurant visitors eat more than non-regular customers this could be a bias in the results. This paper helps to find under which circumstance price does have effect on the amount of food eaten in an all-you-can-eat restaurant and thus helps to narrow down when the sunk cost fallacy exists.
Bibliography
Arkes, Hal R., and Peter Ayton. "The sunk cost and Concorde effects: Are humans less rational than lower animals?." Psychological Bulletin 125, no. 5 (1999): 591.
Arkes, Hal R., and Catherine Blumer. "The psychology of sunk cost."Organizational behavior and
human decision processes35, no. 1 (1985): 124-140.
Brooks, Z., March 2007, The M.L. Guide to Beating the All You Can Eat Chinese Food Buffet
Frank, R. H., and B. Bernanke. Principles of Microeconomics. 3rd ed. New York: McGraw-Hill, 2006. Friedman, Daniel, Kai Pommerenke, Rajan Lukose, Garrett Milam, and Bernardo A. Huberman. "Searching for the sunk cost fallacy."Experimental Economics10, no. 1 (2007): 79-104.
Garland, Howard. "Throwing good money after bad: The effect of sunk costs on the decision to esculate commitment to an ongoing project."Journal of Applied Psychology75, no. 6 (1990): 728.
Heckman, James J. Sample selection bias as a specification error.Econometrica: Journal of the
econometric society, 1979, 153-161.
Just, David R., and Brian Wansink. "The flat-rate pricing paradox: conflicting effects of “all-you-can-eat” buffet pricing."The Review of Economics and Statistics93, no. 1 (2011): 193-200.
Khan, Bashir A., Stephen B. Salter, and David J. Sharp. "Pouring good money after bad: a comparison of Asian and developed country managerial decision-making in sunk cost situations in financial institutions." Asia Pacific Development Journal 7, no. 2 (2000): 105-120.
Kuczmarski, Marie Fanelli, Robert J. Kuczmarski, and Matthew Najjar. "Effects of age on validity of self-reported height, weight, and body mass index: findings from the Third National Health and Nutrition Examination Survey, 1988–1994."Journal of the American Dietetic Association 101, no. 1 (2001): 28-34.
Levitsky, David A., Craig A. Halbmaier, and Gordana Mrdjenovic. "The freshman weight gain: a model for the study of the epidemic of obesity." International journal of obesity 28, no. 11 (2004): 1435-1442.
Mankiw, N. G. Principles of Microeconomics. 3rd ed. Mason, OH: Thomson South-Western, 2004.
McAfee, R. Preston, Hugo M. Mialon, and Sue H. Mialon. "Do sunk costs matter?." Economic
Inquiry 48, no. 2 (2010): 323-336.
Siniver, Erez, Yosef Mealem, and Gideon Yaniv. "Overeating in all-you-can-eat buffet: paying before versus paying after." Applied Economics 45, no. 35 (2013): 4940-4948.
Siniver, Erez and Gideon Yaniv. "All-You-Can-Eat Buffet: Entry Price, the Fat Tax and Meal Cessation." The BE Journal of Economic Analysis & Policy 12, no. 1 (2012): 1-21.
Roberts, A., May 2004, All You Can Eat Salad and Pizza at The Patio
Staw, Barry M. "Knee-deep in the big muddy: A study of escalating commitment to a chosen course of action." Organizational behavior and human performance 16, no. 1 (1976): 27-44.
Staw, Barry M. "The escalation of commitment to a course of action." Academy of management
Review 6, no. 4 (1981): 577-587.
Strough, JoNell, Clare M. Mehta, Joseph P. McFall, and Kelly L. Schuller. "Are older adults less subject to the sunk-cost fallacy than younger adults?."Psychological Science 19, no. 7 (2008): 650-652. Thaler, Richard H. Mental accounting matters. Russell Sage Foundation. Princeton, NJ: Princeton University Press, 2004.
Thaler, Richard. "Toward a positive theory of consumer choice." Journal of Economic Behavior &
Organization 1, no. 1 (1980): 39-60.
Tversky, Amos, and Daniel Kahneman. "The framing of decisions and the psychology of choice." Science 211, no. 4481 (1981): 453-458.
Wang, Youfa, May A. Beydoun, Lan Liang, Benjamin Caballero, and Shiriki K. Kumanyika. "Will all Americans become overweight or obese? Estimating the progression and cost of the US obesity epidemic." Obesity 16, no. 10 (2008): 2323-2330.
Wansink, Brian, and Collin R. Payne. "Eating behavior and obesity at Chinese buffets." Obesity 16, no. 8 (2008): 1957-1960.
Wansink, Brian, and Mitsuru Shimizu. "Eating Behaviors and the Number of Buffet Trips." American
journal of preventive medicine 44, no. 4 (2013): e49-e50.
Whyte, Glen. "Escalating commitment in individual and group decision making: A prospect theory approach." Organizational behavior and human decision processes 54, no. 3 (1993): 430-455.
Appendix Figure 1 Correlations avera ge age thickness scale 1-4 average age Pearson Correlation 1 ,508** Sig. (2-tailed) ,000 N 59 59 thickness scale 1-4 Pearson Correlation ,508** 1 Sig. (2-tailed) ,000 N 59 60
**. Correlation is significant at the 0.01 level (2-tailed).
Figure 2
Figure 3 Correlations average age average number of dishes average age Pearson Correlation 1 ,015 Sig. (2-tailed) ,910 N 59 59 average number of dishes Pearson Correlation ,015 1 Sig. (2-tailed) ,910 N 59 61 Figure 4 Correlations average number of dishes %male average number of dishes Pearson Correlation 1 ,244 Sig. (2-tailed) ,061 N 61 60 %male Pearson Correlation ,244 1 Sig. (2-tailed) ,061 N 60 60
Figure 5
Correlations
perc_sushi perc_misc perc_grill average number of dishes Pearson Correlation ,279* -,112 -,178 Sig. (2-tailed) ,029 ,390 ,170 N 61 61 61 perc_fried Pearson Correlation -,530** ,115 -,023 Sig. (2-tailed) ,000 ,376 ,862 N 61 61 61 perc_sushi Pearson Correlation 1 -,564** -,515** Sig. (2-tailed) ,000 ,000 N 61 61 61 perc_misc Pearson Correlation -,564** 1 ,135 Sig. (2-tailed) ,000 ,301 N 61 61 61 perc_grill Pearson Correlation -,515** ,135 1 Sig. (2-tailed) ,000 ,301 N 61 61 61
**. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).
Figure 6
Independent Samples Test
t-test for Equality of Means
t df Sig. (2-tailed) Mean Difference Std. Error Difference 95% Confidence Interval of the Difference Lower Upper perc_misc Equal variances not assumed -,102 53,757 ,919 -,00200 ,01956 -,04122 ,03721 perc_sushi Equal variances not assumed 1,545 55,984 ,128 ,07519 ,04866 -,02229 ,17266 perc_grill Equal variances not assumed -,550 58,579 ,585 -,01719 ,03129 -,07981 ,04542 perc_fried Equal variances not assumed -1,241 53,130 ,220 -,03243 ,02613 -,08483 ,01997 Figure 7 Coefficientsa Model Standardized Coefficients t Sig. Beta 1 (Constant) 3,522 ,001 wed or fri -,082 -,620 ,538 average age ,026 ,198 ,844 %male ,305 2,405 ,020 perc_misc -,051 -,308 ,759 perc_sushi -,136 -,544 ,589 perc_grill -,274 -1,621 ,111 perc_fried -,448 -2,584 ,013
