Trusting Surveys

In their new paper, Manipulation of Social Program Eligibility (AEJ: Economic Policy 2011), Adriana Camacho and Emily Conover offer an analysis of Columbian census sampling on election years.  They find that sampling is likely manipulated right before election years to increase the number of beneficiaries of welfare programs.

The lines below denote elections.  One can see a large relative sampling increase of the poor before and during election years.

Rule of thumb: when surveys have consequences, their outcomes will be manipulated.