Polling is a little complicated, but not a complete mystery.
As long as a randomized subset of likely voters is correctly selected, the poll should be accurate to the margin of error. There are few external influences that can make poll results less valid such as mid-term backlash to the ruling party or organized efforts to shift primary voters of one-party to the other (rare, but happened in Illinois in 2014). Also, some demographics under-poll.
Poll results are generally considered a snapshot of the campaign at that moment. As the actual campaign progresses, mailings, TV ads, door knocking, attack ads, and robocalls will all have some effect on the campaign results.
Campaigns that are desperate will frequently release “push polls” or outright false polls. These are designed to act as a political tool to generate free media or momentum, but they are not accurate. This “weaponized” polling gives polling a bad name. It confuses the media who have become skeptical in polling results.
How to determine a good poll from a “push” poll.
- The poll should be in general alignment with other reputable polls.
- The methodology should explain how the subset of voters was selected.
- The cross tabs should proportionally reflect the whole population by gender, age, and geographical distribution.
- The easiest way to spot a push poll is to get access to the question asked in the poll. For example, “knowing candidate A was arrested on a DUI, for which candidate would you vote?”