Researchers randomly selected 1,790 Fort Collins households to receive the survey questionnaire. Of this total number, Survey Sampling, Inc. randomly selected and provided 1,700 household addresses from a sampling frame of listed household addresses within the Fort Collins jurisdictional boundaries. This sampling frame included off-campus housing for Colorado State University (CSU) students. Vantage Marketing Research, Inc. randomly selected an additional 90 addresses from a sampling frame of dormitory housing at CSU, and combined the two samples. Survey questionnaires and cover letters requested that one adult from each selected household complete the questionnaire. To ensure random selection of adult respondents, the cover letters and questionnaires asked each household to select the adult (if more than one in the household) who had most recently had a birthday.
Researchers mailed survey questionnaires to the selected households in late March 2006, and responses were accepted until April 28, 2006. Sample households were contacted up to three times in order to maximize the response rate. Prior to distributing questionnaires, each selected household received a notification card, telling the household of their selection and inviting them to participate. Survey questionnaires followed the cards after three days. After a period of two weeks, researchers mailed a follow-up questionnaire and cover letter to households that had not responded to the first questionnaire mailing. Tracking numbers on the questionnaires prevented second mailings to households that had responded to the first questionnaire mailing and to addresses that were vacant or otherwise undeliverable.
Sample Size, Response and the 95 Percent Confidence Interval
Of the 1,790 questionnaires distributed, 103 questionnaires could not be delivered, leaving a sample of 1,687 eligible households. Adult citizens from 786 of these households responded by returning completed questionnaires, representing a response rate of approximately 47 percent. The maximum sampling margin of error for this sample size, at the 95 percent confidence level, is plus-or-minus 3.5 percentage points (the 95 percent confidence interval).