From UtahPolicy.com:

If your campaign is reasonably well organized with a good group of volunteers, you can put together a phone bank and contact a random sample by calling, say, every 10th name in the database of registered voters (or voters who have voted in the last two elections) in the city or district.

The questions need to be carefully written to avoid bias and the callers need to be carefully instructed to avoid bias. However, even with a real effort to be neutral, a survey done by volunteers will not be completely unbiased and accurate. But if 300 or so good responses are received, the campaign can get a feel for where things stand.

To test issues and messages, a campaign can also put together a focus group with 10 or 12 undecided voters. The person conducting the focus group must be skillful and unbiased, or the discussion and responses will be skewed. But a campaign can get a good feel for issues and concerns that are important to voters in the district. Beware, however, that amateur and volunteer efforts at survey research, conducted badly, can do more harm than good.

Read the Rest…