During a casual evening in San Francisco, you decide to go to the movies and decide to purchase a non-refundable ticket for $10. However, when you are supposed to hand your ticket to the clerk, you frantically search through your pockets and realize you can’t find your ticket. You must have dropped it somewhere. You turn around and walk back outside. What will you do?
To gather data, I surveyed Santa Clara University students using an online Google survey. I guessed that the best way to get a large number of responses would be to reach out directly to as many contacts as I could think of and send individual emails. I contacted approximately 230 individuals that I knew through classes, various clubs and dorms, and finished with 181 responses.
The questionnaire featured four sunk cost fallacy scenario questions, as well as demographic questions (gender, field of study, grade level, etc.) and six questions from the Need for Cognitive Closure (NFCC) scale, which I discuss later. My aim was to test whether students honored sunk costs in the scenario questions, then check to see if year in school, field of study, family income, GPA, sleep, exercise or NFCC score correlated with the outcomes of the scenario questions.
If you're curious, you can take the survey here.
Survey Participant Statistics
Sunk Cost Scenario Results
For the two academic questions, other factors may be at play besides the tendency of an individual to consider sunk costs of time and effort. Despite the explicit wording in the minor question that “you will not gain anything from the minor and would rather take other classes,” students may still believe that the long-term value of getting the minor would outweigh the costs of taking classes they dislike. This line of reasoning could explain why a higher proportion of participants would have completed the major than would have completed the class.
Although the wording of my questions assumed that nothing positive would come out of a decision to continue taking the class or finishing the minor, this is often not the case. If a student was losing motivation to complete their degree, the sunk costs of time and effort they had already invested could have positive value.
As one part of my survey, I did a basic assessment of participants using the Need for Cognitive Closure (NFCC) scale. The NFCC scale measures the motivation of people to find a concrete solution to an ambiguous problem or situation, even if that solution is wrong. Someone who scores high on the NFCC scale strongly favors certainty, order, structure, concrete solutions and agreement. A low score on the scale would mean an individual is more comfortable with open-ended questions and may take longer to decide. This scale was introduced to me by Dr. Brunchmann, social psychology professor at Santa Clara University.
My hypothesis was that individuals with high NFCC scores would be more likely to honor sunk costs because this would indicate a fear of the uncertainty associated with quitting and a preference to continue along a course of action.
To analyze this portion of the data, I split the participants into two groups: one with high NFCC average scores and one with low scores. I then compared the averages of the three sunk cost scenario questions (Would you drop the minor? Would you drop the class? Would you buy another ticket?) between the two groups.
The overall average of the low group was only 5% below the average for the high group, and the difference was not nearly statistically significant (p=.62). Even when only comparing answers for the ticket questions (which could be considered the purest sunk cost scenario questions) the difference in the means of the two groups could easily be explained by chance (p=.51).
The most likely explanation for the lack of correlation in these results is that both the sunk cost survey questions and the NFCC scale were oversimplified in my survey. The academic questions bring in an individual’s attitude and beliefs about classes and minors, which do not necessarily relate the sunk cost fallacy. Additionally, the NFCC scale tests for five variables, not all of which were represented on my survey, and not all of which may relate the sunk cost fallacy.
Need For Cognitive Closure Scale Findings
To test for correlation between specific variables and decision-making, I stratified the large group of students into smaller groups based on a trait, then tested the difference between means for the sample groups. The mean considered the two academic scenario questions and the second ticket scenario, in which the loss of $10 is framed as a sunk cost. Overall, students honored sunk costs for the three scenarios 55% of the time.
I began by splitting the population into four groups based on grade level. Although my sample was comprised of over 50% first years, each group still had at least 17 students, so I felt confident testing for correlation. Based on my previous research, I knew that older adults were less likely to honor sunk costs in their decisions, but I did not expect to see a statistically significant difference between grade levels.
I used a linear regression to compare the grade level of each group with the average percent of the time they honored sunk costs.
What traits correlated with better decisions?
You have completed all but two classes for a second minor. You have room in your schedule to take the classes, but you will not gain anything from the minor and would rather take other classes. What would you do?
Four weeks into the quarter, you find out that your philosophy class does not count towards your major. Dropping the class will not hurt your record. But, you have already spent about five hours each week on homework and just took the first midterm. You think you would be better off dropping the class, but the quarter is almost halfway over. What do you think you will do?
Finish the minor
Take other classes
Drop the class
Stay in the class
Buy another ticket
Leave the theater
You are spending an afternoon in downtown San Jose for fun. You decide to go to the movies and are about to buy a $10 ticket. However, you then discover that you dropped a $10 bill out of your wallet. You have more money. What would you do?
Buy another ticket
Leave the theater
The purpose of these two questions is to present two different framings of the same situation to test if a higher proportion would choose to leave the theater when the loss of $10 was framed as a sunk cost. A rational decision maker would make the same decision in both situations, because both ask if the student would like to buy another ticket after losing $10. Notice the difference in wording between the two situations. The first scenario acts as a control group to establish a baseline for how many students would leave the theater when faced with any loss of money.
In the charts, the blue corresponds with a tendency to honor sunk costs, therefore representing what can be considered irrational decisions.
24% of students changed their mind between these scenarios.
When the loss of $10 was framed as a sunk cost instead of an unrelated event, nearly a quarter of students switched answers and said they would not buy another ticket. This irrational behavior is strong evidence for the sunk cost fallacy (p<0.001).
Comparison of High and Low NFCC Groups on Percent Honoring Sunk Costs
had a GPA
came from families earning over $200,000 a year
compared with less than 5% of families making over $200,000 nationally.
came from families earning less than $60,000 a year
Nationally, 54% of families earn less than $60,000 a year.
average hours of sleep / night
Although this finding is consistent with prior research, I was surprised to discover such a strong correlation in my survey. One explanation is that because several of the questions related specifically to academics, seniors were much wiser than freshmen about their academic priorities.
Next, I tested to see if any differences existed between grade levels. I checked the means for each of the five major categories: business, engineering, social science, natural science and humanities.
To my surprise, all majors were very consistent, except for engineering students, who displayed a much lower propensity to honor sunk costs. I used a two-sample t-test comparing engineering students to the entire group of students to see if there was a statistically significant difference between the mean likelihoods to honor sunk costs. I ended up with a p-value of .06.
I found a correlation coefficient of -.96, demonstrating a strong negative correlation between grade level and likelihood to honor sunk costs. This means that as grade level increased, students were less likely to honor sunk costs.
Proportion of Students Honoring Sunk Costs by Field of Study
One possible explanation would be that engineers have more experience with a logical problem-solving approach, and may be less likely to let emotions (the pain of wasting time or money) impact their decisions.
It should be noted that social science majors displayed the highest likelihood of honoring sunk costs. Although the difference is small and likely inconsequential, I find it a bit ironic that students who supposedly think about decision-making the most were least likely to answer rationally.
Exercise, sleep, gender and family income did not strongly correlate with likelihood to honor sunk costs, as I expected. Exercise and sleep may be related to stress and mindfulness, which may affect decision-making. However, the imprecision of my survey data made me reluctant to test for a large connection.
Continue to learn more about my conclusions, further questions and limitations of my results.
With my sample size, this means there is a 6% chance that a difference in means as large as what was observed would occur by random chance. This is fairly strong evidence that engineering students are less likely to honor sunk costs.
The large difference in behavior between the two questions confirmed my hypothesis that framing the loss of $10 as a sunk cost would increase the likelihood that students would choose to leave the theater. In the first scenario, the loss of $10 was unfortunate, but this money had not been invested in a ticket. In the second scenario, participants already owned a ticket they had invested money in, so the loss of this ticket was more painful than losing a $10 bill.
The wording of the second question may have also had an impact. It involved a "frantic search" and an interaction with a clerk which could have been more embarrassing and distressing than the first situation.
My other sunk cost fallacy questions presented two academic scenarios.