Note: This experiment tested the use of postcards to boost voter registration. The study does not tell us about the utility of postcards to boost voter turnout. It would be an apples-to-oranges comparison to try to extrapolate any effect of postcards on voter behavior from an experiment about voter registration. We will be planning experiments to test the use of postcards for voter turnout, to see if the encouraging results we’ve seen for voter registration may extend to turnout.
Objective: To determine if the receipt of a handwritten postcard a week before or a week after receipt of an official voter registration form increased the odds of people completing & sending the voter registration form back to their Secretary of State.
Specifics: Sister District Action Network (SDAN) partnered with the Voter Participation Center (VPC) to send handwritten postcards encouraging registration to eligible, unregistered individuals living in Michigan, Pennsylvania, and Arizona in March 2018. These postcards were sent in conjunction with VPC’s Voting Age Population Program, which sent official voter registration forms to these same people.
The major finding of this study was that handwritten postcards sent after the receipt of voter registration materials (chasers) had a significant positive impact on the odds that a person would return their materials. Unregistered individuals had 20% higher odds of returning the completed form when they were sent a chaser, compared to the odds of returning the registration form after receiving the form alone (without a postcard). This finding achieved statistical significance (p=.009), after controlling for individual-level variables.
Sending a chaser was more effective than sending a postcard prior to receiving the registration form (i.e., a primer). But both target groups that received postcards completed their forms in higher numbers than the control group that did not receive any postcard. This means that receiving a postcard increased the odds of registration completion by about 14%. This finding also achieved statistical significance (p=.036), after controlling for individual-level variables.