Mechanics of Polling
Data collection methods
- Computer-assisted telephone interviewing
- Computer-assisted personal interviewing
- Computer-assisted self-administered interviewing
- Internet self-administered interviewing
- Audio computer-assisted self-administered interviewing
- Mail questionnaire
- Secret ballot
- Self-administered questionnaire
- Face-to-face - i.e., interviewer-administered questionnaire
- Standardized interviewing vs.
- Conversationally flexible interviewing
- Face-to-face interviewing (which can cost as much as $1,000.00 per subject)
- Random digit dialing (RDD) telephone interviewing (which can cost $2.50 to $6.00 per minute)
- Internet interviews (though a self-selected sample may not produce a result that can be generalized to a broader population)
- Computer assisted interviewing
A potential source of polling error is in the data collection. The in-person interview was for years the standard method for random sample polling. The lion's share of polling today is done by telephone, because it is much faster and cheaper.
In 1950 only 62% of American homes had a telephone. Conducting a survey at that time entirely with telephone interviews would have omitted more than a third of the nation's households, introducing serious bias. Today, with a telephone in 97% of all households, the chances of bias are significantly reduced. They are not eliminated entirely.
To reduce the problem created by unlisted telephone numbers, most polling organizations now use a technique known as random digit dialing. Computers are used to generate four digit random numbers. These are paired with the three-digit prefixes the polling organizations get from the telephone companies. The number of interviews allotted to each three-digit prefix is determined by the number of telephone numbers assigned to each prefix.