|
Activity Number:
|
285
- New Advances in Sample Design and Adjusting for Survey Nonresponse
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 11, 2021 : 1:30 PM to 3:20 PM
|
|
Sponsor:
|
Section on Statistical Consulting
|
|
Abstract #318421
|
|
|
Title:
|
Sample Design with Operational Constraints for Zero-Inflated Response Data
|
|
Author(s):
|
Chauncey M Dayton* and Mary Batcher and NJ Scheers
|
|
Companies:
|
BDS Data Analytics, LLC and BDS Data Analytics, LLC and BDS Data Analytics, LLC
|
|
Keywords:
|
Sample design;
Zero-inflated data
|
|
Abstract:
|
Random sampling designs are based on established theory available in classic tests such as Cochran or Kish. We design plans in connection with financial disputes where party A claims that payments are due from party B for up to thousands of transactions. These payment-due claims are utilized as auxiliary variables for creating sampling designs and may be truncated in that values below some relatively small dollar amount are not pursued. What makes our sample design situation different from textbook presentations is that the response variable is an adjudicated value of the auxiliary variable and is often $0. Some claims may be denied and from experience we expect this to occur in up to 50% of the sample cases. We consider the implications for sample design of knowing that the response variable is zero-inflated. Although confidentiality prevents using real data, we have created synthetic data that closely resemble those that we encounter in practice. We demonstrate computations in Excel spreadsheets to simulate varying degrees of zero-inflation to assess its impact on standard errors for estimated total dollar amounts owed using mean estimation and related estimators.
|
Authors who are presenting talks have a * after their name.
