Interviews may be conducted either in person or by telephone. Telephone interviews are less expensive to conduct, allow for better sampling designs, and can be used in conjunction with computers. Computer-assisted telephone interviewing (CATI) systems are available and can increase the efficiency of telephone interviews. A CATI system can be used to help the interviewer adjust their questions based on information obtained during the interview, and allows for the direct entry of data as the interview proceeds.
Typically, prior to the main survey mailing, the questionnaire is pretested on a small sample. Any problems that are identified in the construction of the questionnaire can then be corrected.
Several techniques are commonly employed to boost the return rate for mail surveys. These include the use of advance notification, attractive first-class stamps rather than bulk postage, hand addressing, postage-paid return envelopes, carefully-timed reminder postcards, and repeat mailings of the questionnaire to nonrespondents. Token incentives included with the survey are sometimes used to increase the return rate, but these will also increase survey costs. Incentives may also introduce bias into the returns if they tend to motivate some groups more than others.
Good survey design and sampling technique are critical to the success of sample surveys conducted by any method. Care must also be taken in the data collection and entry process, to avoid introducing errors. Finally, even a well-conducted survey will not yield meaningful results if data analysis and interpretation are flawed. Thus, while the concepts behind public polling are reasonably straightforward, there is a fair amount of art and science involved in conducting a useful study. Gross design and execution errors can lead to meaningless or misleading results. More subtle errors may not completely invalidate survey results, but can decrease the reliability of the study.
If you are interested in conducting a public survey but lack the necessary technical background or resources, there are various sources of assistance available. Survey research units are associated with a number of state college and university campuses. Some of these units,can contract with cities or counties to design or conduct surveys. Others may provide information or assist in studies on a cooperative basis. In addition, a number of private firms specialize in conducting public surveys primarily for market research. The scope of services provided and quality of work performed by these or other consulting firms can vary widely, so careful shopping is advised.
For all but the very smallest municipalities, assessments of citizen attitudes and knowledge will be based on polling a representative sample of the total population. While most of the points noted under Sampling from Populations apply, demographic factors also need to be considered to avoid bias in the study design. For instance, Sommer et al (1990) found that compared to younger citizens, older citizens were more likely to have negative opinions about street trees in front of their homes. It may be desirable to account for differences due to age, sex, sociological, or other demographic factors in the survey. Such information may help local governments decide whether education or other programs need to be targeted toward certain segments of the population.