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.
10 Discussion of Annoyance Survey Results Janssen et al. (2011, Table 1) list response rates for selected aircraft noise surveys that have been conducted previously. The most recent surveys in their review were conducted in Europe by mail or telephone, and they report response rates ranging from 52% to 39%. They do not, however, describe how these response rates were calculated, and response rates can vary drastically depending on what units are excluded from the denominator. In addition, response rates often differ between Europe and the United States because many European surveys can take advantage of detailed population registers for a sampling frame, thereby excluding ineligible addresses from the frame. For the Zurich Airport Study described in Brink et al. (2008), for example, population lists were obtained from 56 of the 68 municipalities surrounding the airport, and the response rate was calculated using only the 56 municipalities that provided listings. In each of these municipalities, the survey was mailed to a randomly selected sample of 60 persons who are known from the population lists to be age 18 or over and who have lived at the address at least one year. The reported response rate of 54% is thus achieved by excluding highly mobile persons, and persons who reside in municipalities that decline to provide population lists, from the sampled population. These exclusions mean that the survey does not include the entire adult population of the area, which can result in coverage bias. The Zurich Airport Study could also take advantage of the detailed information in the population lists by addressing the surveys to the desired sampled individuals, with a personalized approach that can increase response rates; mail surveys in the United States can only be sent to the selected address. AAPOR (2011) provides standard definitions for calculating of response rates, and outlines how results can vary depending on the definition used. For the ACRP Study, a very conservative response rate was calculated, in which only the postal non-deliverables were subtracted from the denominator for both the mail and telephone surveys. This was done so that both surveys would use comparable denominators for the response rates. Recent research indicates that a low response rate is not necessarily indicative of nonresponse bias. Langer (2003) provides an overview of issues involved in interpreting response rates. Groves (2006), reviewing 30 studies in which nonresponse bias was investigated, found no relationship between response rate and nonresponse bias. In an airport survey, the estimate of a dose-response relationship would exhibit nonresponse bias if nonrespondents were systematically more (or less) likely to report annoyance than respondents to the survey with the same noise exposure and demographic characteristics. This cannot be determined from the ACRP Study: to evaluate whether nonrespondents are more (or less) annoyed than respondents it would be necessary to obtain responses from a random subsample of the nonrespondents. TNO and RIVM (1998) followed up with a small subset of nonrespondents to their 1996 survey around Amsterdam Schiphol airport, and found that nonrespondents reported less annoyance to aircraft noise and were more likely to have lower educational levels and to be members of ethnic minority groups. In the ACRP Study, there is evidence that respondents differ from the general population in the airport communities surveyed: respondents to both surveys are more likely to be over age 50 and to be white non-Hispanic, although the respondents to the mail survey more closely match the population characteristics given by the 2010 Census. Among the respondents to the survey, however, there is no statistically significant relationship, after accounting for the level of noise exposure and airport-to-airport differences, between these demographic characteristics and annoyance. The ACRP Study itself may be viewed as testing possible nonresponse bias in annoyance, because two modes with different response rates found similar annoyance levels. The analyses presented in Section 8 found no relationship between the survey mode (mail or telephone) and annoyance, even though the response rate to the mail survey is approximately three times as great as that to the telephone survey. It is possible, however, that the nonrespondents to both surveys might differ from the respondents with respect to percent HA. 29
In Section 2.1 , the potential survey modes that could be used for surveying the population in an airport community about reactions to airport noise were rated with respect to cost, suitability, and possible bias due to undercoverage, nonresponse, respondent selection, and other sources. In Table 1, the mail survey mode was ranked above the telephone survey mode on response rate and coverage, but below the telephone survey mode on respondent selection and follow-up. The results of the ACRP Study show that the mail survey mode is indeed superior to the telephone survey mode on response rate and coverage, with a response rate approximately three times that of the telephone survey. The potential concerns about respondent selection bias in a mail survey do not appear to be borne out in this study. Overall, it appears that in the majority of households, the adult with the next birthday filled out the mail questionnaire as requested. In addition, there is no evidence that there is a difference in annoyance between respondents to the mail survey and respondents to the telephone survey. Although both modes have a higher percentage of older adult and white non-Hispanic respondents than reported by the Census for the areas, the mail survey is closer to the Census figures on these demographic variables than the telephone survey. A mail survey does not obtain response rates that are as high as those from an in-person survey. If a mail survey were adopted for a national airport and annoyance survey, however, it is possible that the response rates could be increased from the 35.1% observed in the ACRP Study by varying incentive amounts, employing a Spanish-language questionnaire, making use of Express Mail for follow-ups, and other means (Montaquila et al., 2010). 30