Telephone canvassing has been one of the
most widely used voter mobilization strategies. Yet in their
ground-breaking study of a field experiment, Gerber and Green (2000)
found that phone calls encouraging people to vote reduce
turnout by five percentage points on average. In this article, I
introduce statistical methods that enable us to uncover discrepancies
between designed experiments and actual implementation. Application
of this methodology to Gerber and Green's data shows that their
negative finding about get-out-the-vote calls is caused by inadvertent
deviations from their stated experimental protocol. I adjust for these
implementation errors using a more appropriate statistical method, and
demonstrate that telephone canvassing increases voter turnout
by about five percentage points on average. This article shows that
statistical methods are essential for detecting and correcting
complications that commonly arise in field experiments. (Note: This
paper was submitted to APSR in August 2002, and was accepted for
publication in August 2003. However, its actual publication has been
delayed because the original authors have not finished writing their
response.)
|