Friday, January 7, 2011

Is there really no selection bias in laboratory experiments?

Whenever you read about a survey or an experiment, the first worry one should have is whether there is some selection bias in the studied example. As I have argued before, experimental economics is almost exclusively on a sample from a minority of the world population. But assuming that we are only interested in this minority (and unfortunately we are), is there still some selection bias.

Blair Cleave, Nikos Nikiforakis and Robert Slonim did some experiments in the classroom with over 1000 students, and then invited them for more experiments in a laboratory setting. Those that followed the experiment did not have different characteristics, which is reassuring. However, this only partially alleviates my worries. Indeed, students are only a small minority of the current population, one that is more educated, coming from a richer background, younger, etc. I am looking forward to a broader study...

1 comment:

Kansan said...

I am much more worried about the replicability of experiments. It is so easy to manipulate the results...