The experiment is a combination of four crossed interventions: (i) flyer treatments, (ii) payment and duration of the survey, (iii) survey content announcement at the door, and (iv) incentives to claim non-voter status. One of five different flyers were left at each house, informing the household about the next day's survey: No Flyer, Survey Flyer, Election Flyer, Opt-Out Flyer, and Election Opt-Out Flyer. Survey flyers simply stated the household would be asked to complete a survey. Election flyers specified that the survey would be about 'your voter participation in the 2010 congressional election." Households in the Opt-Out Flyer treatment receive a flyer as in the Survey Flyer treatment, except for an added check-box which the household can mark if it does not wish to be disturbed. Similarly, the flyer in the Election Opt-Out Flyer treatment has an added opt-out check box.
Second, payment ($5 or $10) and the time necessary (5 or 10 minutes) for the survey were randomly varied and pre-announced on flyers. Lastly, half the respondents of the ten-minute survey were informed that the survey would be eight minutes shorter if they stated that they did not vote in the 2010 congressional election. For voters, this treatment amounts to an incentive to lie and permits us to quantify the disutility of voters were they to say (untruthfully) that they did not vote. For the 50 percent of non-voters who lie without such incentives, this treatment provides an incentive to tell the truth.
In a secondary experiment, in the five days before the election date, researchers posted a flyer on the doorknob of households informing them that 'researchers will contact you within three weeks of the election [...] to conduct a survey on your voter participation.'