Chris Blattman

Search
Close this search box.

Does Michelle Obama ruin Halloween for kids? Dean Karlan says yes.

OK, so Dean Karlan did not actually say that. I might be putting words in his mouth.

What does Dean say? Those of you who know him know that no randomized trial opportunity is ever missed–least of all the arrival of hundreds of kids on their doorstep every October 31.

Enter the infamous Hallow’s Eve experiments.

This year’s question: Does being randomly assigned to Michele Obama (vs Ann Romney) inspire children to choose fruit over candy?

20% of those randomly assigned to the Ann Romney treatment opted for fruit over candy. But 36% of those randomly assigned to the Michelle Obama treatment opted for fruit rather than candy. Nothing was said beyond “what do you want, fruit or candy?” The treatment was simply having a big photo of Michele (or Ann) staring at you as you made your choice.

No word on whether the photo included Michelle’s arms.

(And no he did not tell me the standard errors. I say this because I know my audience is nerdy enough that I will get 14 questions about this if I don’t mention it.)

My conclusions:

First, I apparently need a Michelle Obama picture in my apartment, because tonight I ate an inordinate number of peanut butter cups and rasinettes.

Second, as far as I can tell, it actually sounds like it is Dean Karlan who ruins Halloween for kids. But we parents deserve at least one house to enjoy, so I’ll take Dean’s. Unless, of course, Chris Udry is offering Vodka shooters on his porch across the street.

17 Responses

  1. well, you would need a sample size of more than 120 to get a 2 sigma significance : – )

    just enjoying to read Deidre McCloskey: The cult of statistical significance

Why We Fight - Book Cover
Subscribe to Blog