1994 Citizen Attitude Survey__Report of Findings

Home Browse Search

households. As this was the mailing list used to develop the random sample to be surveyed, the 5,797 household figure will be used in this report. Louisville Households Surveys Mailed Returned Surveys Table 2 HOUSEHOLD SAMPLE 5,797 1,080 345 100 % 18 % 6 % Each mailing of the survey instrument was developed by generating a random numbers table and assigning each number to a corresponding address from the list of Louisville households. The second mailing did not include the first mailings 570 addresses. The second random numbers table only pulled 510 addresses from a list that went from one (1) to 5,287, not the full range of the 5,797 households in Louisville. An analysis was conducted to compare the two separate mailings to determine if the two "sample drawings" were indeed representative of the same population. This is an indirect measure of whether or not randomization has been violated in the sampling procedure involving two mailings. In addition, it was decided to combine the results of the two mailings if statistical tests did not show appreciable differences in the mean scores of key social -economic indicators. A means test and equality of variance test was conducted on the distribution of the two samples; by age, by years of residence in Louisville, by annual household income, and by number of people living in the household. Those test results are displayed in Table 3. Table 3 COMPARISON OF TWO MAILING POPULATIONS Item Age Years of Residence (Louisville) Annual Household Income Number of Persons in Household Mean of Mean of Two -Tailed Means Equality of First Second Test (.05) Variance Test Mailing Mailing (.05) 42.1 42.4 .82 .64 9.1 10.5 .27 .02 $60,514 $59,977 .89 .90 2.9 2.9 .67 .54 This analysis shows very little difference in the mean values and variance for the two mailings. The tests were not significant. We may be reasonably certain that respondents in the two 6