Activity Number:
|
284
|
Type:
|
Contributed
|
Date/Time:
|
Tuesday, August 4, 2009 : 8:30 AM to 10:20 AM
|
Sponsor:
|
Social Statistics Section
|
Abstract - #305514 |
Title:
|
Confidence Intervals for Radio Ratings Estimators
|
Author(s):
|
Richard Griffiths*+
|
Companies:
|
Arbitron, Inc.
|
Address:
|
9705 Patuxent Woods Circle, Columbia, MD, 21046,
|
Keywords:
|
Confidence Intervals ; Wald ; Clopper-Pearson ; Coverage Probability ; Radio Ratings
|
Abstract:
|
Arbitron's current method for forming confidence intervals for radio rating estimates is based on the normal distribution. It is well-documented that with estimates of proportions from sample surveys the usual confidence interval method based on the normal distribution is generally lacking in some important situations. One of those situations is when the estimated proportions are very small and especially when sample sizes are also small. Because Arbitron ratings represent estimates of small proportions, we recently re-evaluated our confidence interval methodology, which was first implemented in the early 1980's when radio listening was less fragmented - ratings were generally larger - and the methodology was less likely to fail. This poster displays the results of an empirical study designed to compare the currently-used Wald confidence intervals to some alternatives.
|
- The address information is for the authors that have a + after their name.
- Authors who are presenting talks have a * after their name.
Back to the full JSM 2009 program |