Economists Do It With Models

Warning: “graphic” content…

Bookmark and Share
How Behavioral Economists Put The “Trick” In “Trick Or Treat”…

November 6th, 2013 · 3 Comments
Behavioral Econ · Decision Making

This Halloween, I did what any sane economist would do on this fine day- I drove to New Haven to experiment on trick-or-treaters with Dean Karlan. When I explained this to my students, I mentioned that back when I was little parents were super afraid of people putting razor blades and other unsafe objects into kids’ candy (Dear Mom: This was apparently an overblown concern), whereas now it seems like one of the biggest things that parents have to watch out for is this:

Yes, that is Dan Ariely in a bee costume. You’re welcome. And it’s particularly thematically relevant, in fact, since Ariely conducted one of the experiments that I describe below. Anyway, it’s pretty typical when conducting research to write up a literature review that summarizes previous work related to the question your research is trying to answer. In addition, sometimes a literature review also touches on previous work that has been done using similar empirical methods, even if the topic of that work doesn’t perfectly match with the current research question. Therefore, I took it upon myself to find out what existing economic research has been done on trick-or-treaters. Obviously. (This is actually somewhat difficult to do, since if I search EconLit for “Halloween” I only get papers about the Halloween effect in stock returns.)

Study 1: Daniel Read and George Loewenstein (1995), “Diversification Bias: Explaining the Discrepancy in Variety Seeking Between Combined and Separated Choices”

Overall, this paper investigates whether people exhibit a diversification bias (i.e. choose more variety for the sake of variety) when they are asked to choose ahead of time what they want to consume in several future time periods compared to when they are asked sequentially at each point of consumption. In order to test whether differences in choice bracketing drives the difference in preferences, the researchers have one group of trick or treaters choose two candies (from a large selection of 3 Musketeers and Milky Ways I think) at one house and another group choose one candy at one house at one candy at a second house. Even though the difference in framing is trivial, since both candies go into the pillowcase or plastic pumpkin for later consumption, all of the children who were asked to pick two candies at once chose one of each kind, whereas only about half of those were were asked to make two separate choices did so.

Study 2: Kristina Shampanier, Nina Mazur, and Dan Ariely (2007), “Zero as a Special Price: The True Value of Free Products”

In this paper, Ariely (of the aforementioned bee costume) and his coauthors examine whether making a product “free” in an explicit sense (but not in an opportunity cost sense) biases consumers towards consumption of the free product, even when they have to pass up a better deal in order to get the free item. To answer the question, children were presented with three Hershey’s kisses and then asked whether they want to trade the kisses for a fun size or a regular size Snickers. In one condition, children were told that the fun size Snickers is free (FREE!) whereas they would have to trade in a kiss for the regular Snickers. In a second condition, children were told that they could either trade one kiss for a fun size Snickers or two kisses for a regular Snickers. Even though, in each case, the question to consider is whether it’s worth it to give up one more kiss in order to get the regular Snickers (and I think we can all agree the the regular Snickers is the better deal), far more children opted for the fun size Snickers when it was FREE!

Study 3: Santosh Anagol, Sheree Bennett, Gharad Bryan, Tiffany Davenport, Nancy Hite, Dean Karlan, Paul Lagunes, and Margaret McConnell (2008), “There’s Something About Ambiguity”

This study shows the correlation between ambiguity aversion (being biased against gambles where the relevant odds are not salient) and a child’s choice of costume. In this study, the researchers find that children who won’t take an ambiguous gamble over a clear one (as measured by a quick choice experiment) are more likely to have chosen costumes that are more conventional and commonplace.

Actually, this last one is but the first in a series of fun Halloween experiments that Dean Karlan has done, all of which are helpfully cataloged here. My personal favorite is the one that shows how kids’ preferences for fruit over candy can be strengthened by showing them a picture of Michelle Obama.

Anyway, I feel like I don’t need to write a whole lot about this year’s experiment, since a reporter for the LA Times wrote more than enough about it here. That said, you may have noticed that the article was published on Halloween, so there are no experimental results to be found. In other words, we hadn’t done this yet:

Empirically, here’s what Maria (my RA, pictured above) and I have got so far:

  • Children with parents who made rules about how and when they could eat their candy were more likely to choose fruit as their treat. Put another way, children who chose fruit were more likely to have rules about how they could eat their candy. We hypothesize that this is because the fruit is not subject to the same consumption restrictions as the candy, and this adds to its appeal.
  • Parents are more likely to make rules for children that eat their candy faster than they intend to. (This was measured by comparing how long children expected their candy to last this year to how long they reported it lasting the previous year.)
  • Strangely enough, children whose answers indicated that they had defied their parents’ rules were more likely to choose fruit as their treat.
  • The correlation between whether a child planned his costume well in advance and whether he was going to be organized about how to eat his candy was too noisy to establish statistical significance. (See the illustrative caveat about the logistics of the study below.)
  • For some reason that I can’t intuitively explain, children who went with their first choice of costume were more likely to choose candy over fruit, even when controlling for other observable factors.
  • Also for some reason that perplexes me, children who were surveyed by either Dean or myself were significantly more likely to choose candy than the other kids, so I guess one important thing that we learned is that we are inadvertent candy pushers.

It’s probably not surprising that trying to elicit coherent and accurate responses from young children is pretty challenging, but I think the following photo underscores how much the Halloween element adds to the difficulty:

halloween box

Yep, that’s a Yale undergraduate trying to get answers from a kid in a box. Research ain’t always glamorous, people.

Tags: Behavioral Econ · Decision Making

3 responses so far ↓

  • 1 Brit Lundell // Nov 6, 2013 at 9:30 pm

    A hypothesis to explain the behavior in the second study is that the second condition implicitly prices a volume of Snickers (fun size = 1 kiss), whereas the first condition does not. Children in particular may feel more comfortable “purchasing” the full size Snickers with that information.

  • 2 Timothy // Nov 6, 2013 at 11:58 pm

    I use the Ariely Kisses-Snickers experiment as a lecture on consumer choice, showing that it’s pretty easy to draw budget sets and strictly convex (if a bit odd) preferences such that the same kid could make both choices rationally.

  • 3 Creeper // Nov 8, 2013 at 9:08 am

    That’sssss a very nice sssstatissssstical sssssurvey you’ve got there. It would be a shame if ssssomething were to happen to it.

Leave a Comment