A Quick Guide to Polling: Part 3
Selecting a Pollster
Here is Part 3 on polling. Part 1 covers basic polling terms and you can read it by clicking here. Part 2 deals with the most common types of polls Benchmark and Tracking and you can read it by clicking here.
In selecting a pollster to work with, one of the biggest thing to consider is the quality of their polling model. A polling model is how they decided what their “sample” will look like, or more simply put who they will poll. What the breakdown between Democrat and Republican will be, men v. women, how to define likely voters and so on.
How did PPP (Public Policy Polling) accurately predict the 2012 Florida Presidential race when most of the other big name pollsters not only predicted a Romney victory, but did so outside the margin of error? Because, PPP’s polling model accurately predicted a larger number of younger and minority voters than the other pollsters.
How can you tell if a pollster’s model is accurate? Well you can’t. Sorry, I know that’s not what you wanted to hear, but the news isn’t all that bad. What you want to look for is a pollster with a generally good track record across the nation and specially a strong record of accuracy in your state or region of the country. Geographic areas are very different to poll and hiring someone with an understanding of the area that you are running it can be very helpful.
Ask for references and CALL the references. A candidate I worked with several years ago hired their pollster before they brought me on board. I had never heard of the pollster, but that is not unusual, there are lots of pollsters. After the first poll came back and it was clearly an amateur effort I asked the candidate why they hired the firm? He showed me a list of prestigious candidates and commercial organizations that they claimed were former clients. I called the first three on the list and they had never heard of the company. The fourth recognized the name as a company they had hired and quickly fired because they didn’t deliver what they had promised. We followed suite and fired them the next day. By the way, that first company said the candidate was up by 34 points. The pollster we later hired showed that we were 15 points down. We used their numbers to base our game plan and won by 7 points.
Another thing you should ask for is to see a sample of their previous work. Your polling data is secret and it should be, but clients often will allow their data to be shared a couple of years after the race is over. This will show you the type of data you will receive.
You also need to find out how accessible the pollster will be to help the campaign interpret the data. If you have a team of experienced staffers and consultants this will be less important, but is it still good to have your pollster available to help you use their data correctly.
Decided upfront the number of benchmark and tracking polls and negotiate for the best possible rate.
Next week “do it yourself polling” and why it’s campaign malpractice…