Dan Ariely wrote Predictably Irrational in 2008, conducting a series of experiments that proved people often think—and act—irrationally when making choices.
In one of Ariely’s experiments, Professor Drazen at MIT asked students to write down the last two digits of their social security numbers (We will call this number Y). He then asked two questions:
- With a simple “yes” or “no,” students were instructed to write whether they would pay $Y for a list of items that included two bottles of wine, a cordless trackball, a cordless keyboard, a design book, and a box of Belgian chocolates.
- Professor Drazen then asked them to write down the maximum they were willing to pay for these items (their bids).
In terms of their bids, Ariely found that yes, social security numbers acted as anchors; students with social security numbers that ended high numbers were willing to pay more for this spread of items.
Ariely dubbed this arbitrary coherence: “Initial prices are largely ‘arbitrary’ and can be influenced by responses to random questions; but once those prices are established in our minds, they shape not only what we are willing to pay for an item, but also how much we are willing to pay for related products.” (page 32)
The question I was left with was whether the first question students were asked, about yes or no, had any impact on this idea of arbitrary coherence. Since Arley does not address the results of the first survey question, I decided to head canvas students at Stern School of Business.
Dinner and A Movie: The Experiment
Objective: To what extent must you emphasize a number for arbitrary coherence to exist? Is mere mention of a number enough for arbitrary coherence, or must a direct question relate to said number?
- Group 1: Control group.
- Group 2: Ask for a birth date then ask how much they’d pay for two separate items.
Hypothesis: If arbitrary coherence exists, there will be a statistically significant difference between group 1 and group 2, based on an independent samples t-test.
Study design: To study this question, I surveyed 40 people, 20 in each group. I asked group 1 how much they would pay for a pizza pie and for a movie. I asked group 2 their birthday (date) and then asked them the pizza and movie question. By asking their birthday, I primed group 2 with a number. If arbitrary coherence exists, and assuming a homogenous random sample, there should be a significant difference between the estimated cost of pizza and movies between groups 1 and 2.
In terms of the pizza price, there was not a statistically significant difference between the groups. A t-test of independent samples was performed and yielded a probability value (p) of 0.79. A p-value less than 0.05 is statistically significant, so there is not a significant difference between the primed group and the unprimed group in terms of pizza price (p>0.05,n.s.).
In terms of the movie price, there was not a statistically significant difference between the groups, as the t-test yielded a value of 0.07. It is worth noting that 0.07 is close to 0.05 so it is what statisticians would call “approaching significance” (p>0.05,n.s.).
Conclusion: While there was not statistical significance in my study, this may not conclusively be the final word on arbitrary coherence. As one can see by the graphs, on the whole the priming groups were higher, even though they were not significant. It may be that with a larger sample size or more precise questions in a more controlled environment, arbitrary coherence could be observed.