Aug
12

Designing my pilot test – August update

I’ve spent the past month or continuing to talk to campaigns about my thesis and designing a plan for pilot testing my research mailers before I actually send them out in November. Basically, I am pilot testing my political campaign mailers for two reasons: to make sure that they have the effect I think they will have, and to ensure that the campaigns that send them out won’t get an unusual amount of negative feedback that might cost them votes.

To do this, I designed a survey in Qualtrics that will allow me to test the mailers using Amazon’s Mechanical Turk interface. Mechanical Turk is a handy online marketplace that allows researchers and others to design “tasks” for completion by Amazon users who work on small jobs (like taking a survey) in exchange for a small payment. Mechanical Turk users who are registered to vote will be able to opt-in to my survey task, where they will be split into two treatment groups and a control group. Hopefully I will be able to use the results of this survey in my thesis to talk about the mechanisms I think are at work in the mailer, something I won’t be able to assess in my main project because I will not be collecting any data on subjects beyond whether or not they turned out to vote.

Designing the pilot testing involved a significant amount of research into how I think my mailers will affect people, followed by how best to measure these effects, which are likely to be small. Once the Student IRB gets back to me regarding my project’s approval, I’ll be able to launch my study on Mechanical Turk and gather data to analyze!