Carl Bialik of the Wall Street Journal takes an unusually even handed look (for someone in the national media) at IVR polling.
I sure would like to meet the dog(s) Ann Selzer knows that are capable of going through and answering a whole telephone poll- that's worthy of front page coverage in and of itself!
One nice thing about the story was that a recording of a sample PPP poll was posted on the WSJ website. If you've ever been curious about the voice of our polls, listen here.
Thanks for the link Tom.
ReplyDeleteI listened to your recording, and I wonder if you could talk about your methodology a bit.
For example, do you randomly rotate the order of the names of candidates, as I believe SurveyUSA does? (Do you have tests to show whether or not this even makes a difference?) Do you somtimes rotate the offices you ask And how about asking for party ID before you ask for voting preference?
When you poll a state like Florida, do you think about asking for Cuban and non-Cuban hispanics to identify themselves? Is your sample big enough to make a crosstab like that valuable?
Thanks.
We don't rotate candidate names. In the early days of PPP the folks who set up our process did some pretty extensive testing and found it didn't matter, and I've done some experimenting with it too and come to the same conclusion.
ReplyDeleteI don't think there would be anything inherently wrong with asking for party ID before voting preference, but we find that the best way to get people into the survey is asking who they support for President first. That seems to be the number one way to make them interested in participating.
We have not given any thought to asking about Hispanic subgroups in Florida, although that seems worth considering.
Thanks.
ReplyDelete