Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
8 Comparison of Respondents by Survey Mode One major item of interest is whether the relationship between noise exposure and annoyance differs in the telephone and mail surveys. Because the set of airports used in this ACRP 02-35 survey was selected purposively for carrying out the mode experiment, the data are not suitable for developing a dose- response relationship between noise exposure and annoyance. The statistics presented here are calculated solely for comparing the telephone and mail respondents surveyed within each airport community. We performed a Cochran-Mantel-Haenszel test to assess whether the odds of being highly annoyed is the same between telephone and mail groups, aggregating the information across the twelve subgroups formed by airport and DNL noise strata combinations. The test statistic is ð2 = 1.1 with p-value > 0.20, showing no significant difference overall between the mail and telephone surveys in percent HA. Table 9 displays the difference in percent HA separately for each DNL noise stratum and airport. Each entry in the table is the percent HA from the mail survey minus the percent HA from the telephone survey for that DNL noise stratum and airport. Standard errors for the individual estimates are given in parentheses, and these standard errors are large because of the small sample sizes of telephone respondents in each stratum. After adjusting the p-values for multiple testing using the Bonferroni method,9 only one of the differences is significant at the 0.05 level: DNL noise stratum 65+ for Airport 3, where the percentage HA on the mail survey was 40 percentage points higher than on the telephone survey. Table 9 Difference in percent HA (mail â telephone) by airport and DNL noise stratum 50-55 55-60 60-65 65-70 70+ Airport 1 -5.0 (6.0) -4.2 (8.2) -2.5 (8.3) Airport 2 -0.6 (7.4) -8.7 (11.2) 14.7 (12.2) 5.3 (11.9) 25.3 (11.3) Airport 3 -7.1 (10.6) -1.0 (14.7) 40.0 (10.5) -2.7 (7.4) NOTE: Standard errors are in parentheses. Boldface values are statistically different from zero at the 0.05 significance level, after making a Bonferroni adjustment for multiple testing. We also looked at possible discrepancies between the telephone and mail surveys on the demographic characteristics of the respondents, again controlling for strata. Table 10 shows the results from Cochran- Mantel-Haenszel tests performed for the percentage of respondents who are male, white non-Hispanic, age 50 or over, and who live in a single-adult household, along with the previously reported test results on percentage highly annoyed. The telephone and mail surveys are highly significantly different with respect to the percentage of persons responding who are age 50 and over, and the percentage of responding households that have a single adult. 9 When a significance level of 0.05 is used, one expects one out of every 20 hypothesis tests performed to be statistically significant even if all null hypotheses are true. Multiple comparisons procedures adjust the p-values for the number of tests performed to protect against possible âdata snooping,â in which many hypothesis tests are performed and only the tests with significant results are explored further; see Oehlert (2000, chapter 5). 26
Table 10 Cochran-Mantel-Haenszel test statistics and p-values for testing difference between telephone and mail survey results Characteristic Statistic p-value Percent Highly Annoyed 1.12 0.29 Percent Male 2.92 0.09 Percent White non-Hispanic 1.45 0.23 Percent Age 50 and Over 14.97 < 0.0001 Percent with One Adult in Household 8.00 0.005 Table 11explores these demographic differences in more detail. Each entry in the table is the estimated percentage from the mail survey minus the estimated percentage from the telephone survey. The mail survey has a lower percentage of respondents age 50 or over for every airport and DNL noise stratum. Combining the results across strata using the Cochran-Mantel-Haenszel test gives a highly significant difference between the two survey modes.10 We saw in Section 7 that the mail survey estimates were closer to the Census figures on percentage of respondents age 50 or over. The mail survey also has a significantly lower percentage of respondents that report having one adult in the household. Table 11 Difference in demographic characteristics (mail â telephone) by DNL noise stratum 50-55 55-60 60-65 65-70 70+ Percent White Non- Hispanic Airport 1 --0.5 (5.4) 2.3 (6.6) 5.0 (4.6) Airport 2 -11.6 (11.3) -20.7 (12.5) -19.3 (10.7) 4.0 (9.5) 16.4 (8.1) Airport 3 -6.8 (11.9) -3.4 (12.2) -6.2 (7.6) -6.0 (5.0) Percent Male Airport 1 1.4 (8.4) 5.7 (8.5) 10.9 (8.2) Airport 2 -7.3 (11.5) 21.3 (12.9) 12.3 (12.1) 19.5 (11.7) -6.1 (11.7) Airport 3 -4.8 (13.3) -8.9 (14.7) 11.5 (11.6) 2.3 (8.0) Percent Age 50 and Over Airport 1 -18.1 (8.1) -5.5 (7.9) -10.5 (7.0) Airport 2 -1.5 (11.5) -19.0 (13.0) -17.7 (12.0) -29.4 (11.7) -2.7 (11.7) Airport 3 -0.1 (13.2) -12.2 (14.9) -9.3 (10.9) -9.8 (7.9) Percent in Single-adult Households Airport 1 3.7 (8.1) -7.9 (8.7) -5.2 (8.3) Airport 2 -14.8 (11.4) -12.8 (13.0) -15.3 (12.0) -11.7 (11.7) 4.0 (11.6) Airport 3 0.0 (13.6) -10.9 (14.5) -6.6 (11.0) -22.5 (7.7) NOTE: Standard errors are in parentheses. 10As for the analysis of individual DNL noise stratum differences shown in Table 11, a Bonferroni adjustment should be made for multiple testing if it is desired to examine the difference for a individually. The standard errors given in Table 11 are not adjusted for multiple comparisons. In Table 11, the only individual stratum percentage that is statistically significant at the 0.05 level is the percentage in single-adult households in the 65-70DNL noise stratum in Airport 3. 27
