|
Abstract:
|
Each year Nielsen Media Research (NMR) uses a RDD frame to sample respondent households for its national mixed mode diary surveys of television viewing in the United States. In particular, four times each year (January/February, March/April, June/July, and October/November) NMR conducts a survey of television viewing that begins with a RDD stage. The frame used is all possible 100-banks of telephone number area code/prefix/suffix combinations with at least one number listed in the bank. In each of these surveys upwards of one million RDD telephone numbers are randomly split into four replicates; collectively the four replicates are used for the survey's sampling pool. Following fixed calling rules, which space up to 15 call attempts per telephone number over a 14-16 day field period, NMR makes approximately eight million dialings during each survey. Callbacks are spaced across different days of the week and times of day. This effort leads to far better than average response rates compared to most commercial survey efforts. In the Oct/Nov 2003 survey, for example, an AAPOR Response Rate 1 of 43% and an AAPOR Response Rate 3 of 47% were achieved. This paper will present analyses using NMR's 2003 calling data to illuminate where efficiencies are likely to be gained by changing the fixed calling rules NMR now uses to more flexible calling rules. Past research along these lines suggested that efficiencies may be gained, while keeping the total number of dialings fairly constant, if fewer than 15 call attempts were made to certain call history patterns and more than 15 call attempts were made to certain other call history patterns. We will investigate these findings and present means by which efficiencies may be gained using a more flexible set of calling rules.
|
