Two Polls are Better Than One

As I said in my first post on this subject, there  are times when an underfunded local campaign is better off doing two or more simple polls rather than one large, more complicated and more expensive poll.   Sometimes, this is because knowing how a particular message is doing with the electorate over time is extremely important; sometimes it’s just that polls can be used as fundraising tools.

In 2006,  I ran a campaign for Supervisor in northern Sonoma County.   The campaign had three major problems:

  • We were running against an entrenched incumbent.
  • My client was under treatment for a serious illness throughout the campaign.
  • The campaign did not get underway until mid-January, less than six months before the June primary election which would be decisive, since there were only two candidates.

[Now a small digression:  the margin of error on a sample size of 100 randomly-selected voters is roughly +or – 10%,   My experience working with volunteer poll takers is that 100 completed calls within a window of 3 to 4 days is about the practical limit (and even then I usually had to supplement the volunteers with a few paid callers.)  This is the kind of poll I am referring to when I talk about the 2006 campaign.]

When you have reports coming in from all parts of a district  every day ( and we did), you develop a pretty good rough sense of how you are doing.  With about six weeks remaining, we became pretty sure that we were rapidly closing the gap on the incumbent.   The problem was,  we were also rapidly running out of money, and if things didn’t improve, we might not be able to afford all the voter communication we had planned.

The solution: a 100 likely voters poll every week for four weeks, at a total cost of around a thousand dollars.   We didn’t poll to find out how well our message was working, or to find out how best to attack our opponent.   Instead, our goal was to develop proof that we were closing the gap and to use that information in a fundraising push.  The plan worked beautifully – the poll results were just as we expected, and our candidate was able to use the information in fundraising calls.  In the end, we lost by around 200 votes out of more than fifty thousand cast.


Polling for Local Elections

When I was a political consultant, I spent a lot of time trying to provide the benefits of solid polling data to underfunded local campaigns like school board and city council elections.   I had studied statistics in Graduate School, and I was sure I had the necessary mathematical background to conduct a poll.   Unfortunately, I didn’t at first appreciate the importance of the way questions were formulated to the process.   When I began to write my first poll, I had to turn to my new friends in the Political Science Department at Sonoma State for help in writing them and for advice in interpreting the results.   Professors Don Dixon and John Kramer , among others, were generous with their time, and in 1986, with the help of volunteers like Anne Seeley, I was able to conduct my first poll, an exit poll focused on the Santa Rosa City Council Election.   I continue to provide polling advice to local candidates, even after my retirement.

With hindsight, I can see that I spent too much time over the years in the process of polling, and not enough time helping the candidates I worked with understand how to interpret the information we developed.   For future candidates, I have developed  a short list of key points to remember.   I will elaborate on each of these points in future posts.

  • A poll is a snapshot; it is not a way to predict who is going to win
  • Often two simple polls, properly spaced, are more useful than one long poll
  • Undecided voters are usually much more important to look at than those voters who have made up their minds….but not always
  • You’ve got to know the territory; the context of the poll can be as important as the results themselves
  • Likely voters are selected for polling based upon educated guesses