Polling Data
We have managed over 200 polls since 2001. We no longer conduct polls, but we continue to supply the best data available for polling.
Bear in mind, the data used for polling is processed different than the data used for advocacy calling.
While there are a number of significant issues critical to conducting an accurate poll, one of the most critical issues is that the call list used for the poll must be based on a truly random sample of the universe in which we are interested. This is no easy chore, even with today's advanced database management programs.
We suggest that you review the white paper entitled "Polling Primer" for suggestions and tips on dealing with polls and their results.
Common Criteria
Pollsters use a variety of techniques to identify the universe they are targeting. For political campaigns, there seems to be two popular approaches to targeting likely voters. The first approach involves randomly calling everyone in the community and determining if they are a voter by simply asking the person called if he or she is voter and will they vote in the next election. The second approach, and the one we advocate, is to call only registered voter's households and conduct the interview with the same qualifying questions.
Our recommendation for any voter universe is to randomly select a sample of one voter per household, quality 3 or better, with a telephone.
Obviously, there may be districts or universes where you would need to decrease the quality ranking of the voter sample in order to get adequate sample numbers for the number of completed surveys required.
Sample Size
Sample size is determined by the number of required completed surveys. Pollsters frequently request a sample size that is 15 times the number of surveys needed to complete a poll. For 300 surveys, the sample size would be 4,500 unique telephone numbers, and for 600 surveys, the sample size would be 9,000 unique telephone numbers.
Because of our telephone appending techniques, we have always gotten the required completed surveys using a multiple of 10 rather than 15. This is an important issue when you may have only 3000 or so voters with phones in a small district to accomplish a 300 survey poll. Our data frequently works on smaller districts.
This said, if there are sufficient numbers available, we will always supply a multiple of 15 for the required completed surveys.
Fields Provided
Cross-tab fields are essential to the polling process, whether you collect the cross-tab data yourself by asking questions, or you use the data provided in the sample. The following download indicates the fields we usually send with the data. If you require a particular field, not on the list, please give us a call, and we will accommodate you.
You will note that the voter's name is included, however the address information is omitted. Polling best practices generally dictates that neither the name nor the address be used when listing telephone numbers for polling. Several pollsters have requested the name of the voter, so I generally include it in the data.
We can custom tailor the field list to your specific needs and even produce call sheets if you are doing an in-house poll.