Citation

Leveraging Probability Samples for Better Estimates from Non-Probability Samples

Abstract | Word Stems | Keywords | Association | Citation | Similar Titles



Abstract:

Researchers commonly reweight responses from non-probability samples to match the demographic characteristics of a given target population. When nonresponse is ignorable conditional on demographics, this procedure allows researchers to generate good estimates of population quantities. However, reweighting with respect to demographics is often insufficient and a larger conditioning set that includes non-demographic information is required. In this paper, we propose the “pseudo-probability survey design,” which involves: (1) fielding a subset of questions from a known, benchmark probability sample on a non-probability sample; and (2) reweighting the resulting data to match the demographics and responses of the joint population distribution of the substantive questions of interest. We demonstrate that the availability of a larger conditioning set can make inferences drawn from a non-probability sample more credible. We discuss well-known results on nonparametric identification, estimation, inference, and consequences of misspecification. We illustrate this approach using an augmented replication of a subset of questions asked in the 2016 American National Election Studies survey (ANES), performed on Mechanical Turk in concurrence with the ANES survey. The present study, which we call a pseudo-ANES, considers the empirical implications of conditioning sets composed of demographics, general self-reported political attitudes, and subject-specific attitudes in the domain of gun rights. Full results illustrating whether or not our pseudo-ANES can recover ANES benchmarks will be presented pending the release of the 2016 ANES data, according to our preanalysis plan.
Convention
All Academic Convention makes running your annual conference simple and cost effective. It is your online solution for abstract management, peer review, and scheduling for your annual meeting or convention.
Submission - Custom fields, multiple submission types, tracks, audio visual, multiple upload formats, automatic conversion to pdf.Review - Peer Review, Bulk reviewer assignment, bulk emails, ranking, z-score statistics, and multiple worksheets!
Reports - Many standard and custom reports generated while you wait. Print programs with participant indexes, event grids, and more!Scheduling - Flexible and convenient grid scheduling within rooms and buildings. Conflict checking and advanced filtering.
Communication - Bulk email tools to help your administrators send reminders and responses. Use form letters, a message center, and much more!Management - Search tools, duplicate people management, editing tools, submission transfers, many tools to manage a variety of conference management headaches!
Click here for more information.

Association:
Name: APSA Annual Meeting & Exhibition
URL:
http://www.apsanet.org


Citation:
URL: http://citation.allacademic.com/meta/p1251030_index.html
Direct Link:
HTML Code:

MLA Citation:

Aronow, Peter., Baron, Jonathon. and Coppock, Alexander. "Leveraging Probability Samples for Better Estimates from Non-Probability Samples" Paper presented at the annual meeting of the APSA Annual Meeting & Exhibition, TBA, San Francisco, CA, Aug 31, 2017 <Not Available>. 2018-06-19 <http://citation.allacademic.com/meta/p1251030_index.html>

APA Citation:

Aronow, P. M., Baron, J. and Coppock, A. , 2017-08-31 "Leveraging Probability Samples for Better Estimates from Non-Probability Samples" Paper presented at the annual meeting of the APSA Annual Meeting & Exhibition, TBA, San Francisco, CA <Not Available>. 2018-06-19 from http://citation.allacademic.com/meta/p1251030_index.html

Publication Type: Conference Paper/Unpublished Manuscript
Review Method: Peer Reviewed
Abstract: Researchers commonly reweight responses from non-probability samples to match the demographic characteristics of a given target population. When nonresponse is ignorable conditional on demographics, this procedure allows researchers to generate good estimates of population quantities. However, reweighting with respect to demographics is often insufficient and a larger conditioning set that includes non-demographic information is required. In this paper, we propose the “pseudo-probability survey design,” which involves: (1) fielding a subset of questions from a known, benchmark probability sample on a non-probability sample; and (2) reweighting the resulting data to match the demographics and responses of the joint population distribution of the substantive questions of interest. We demonstrate that the availability of a larger conditioning set can make inferences drawn from a non-probability sample more credible. We discuss well-known results on nonparametric identification, estimation, inference, and consequences of misspecification. We illustrate this approach using an augmented replication of a subset of questions asked in the 2016 American National Election Studies survey (ANES), performed on Mechanical Turk in concurrence with the ANES survey. The present study, which we call a pseudo-ANES, considers the empirical implications of conditioning sets composed of demographics, general self-reported political attitudes, and subject-specific attitudes in the domain of gun rights. Full results illustrating whether or not our pseudo-ANES can recover ANES benchmarks will be presented pending the release of the 2016 ANES data, according to our preanalysis plan.


Similar Titles:
Survey non-response in a national area probability sample as a dimension of survey quality; an analysis of community characteristics

The Consequences of Using Poststratification Estimation to Correct for Differential Non-Response in Sample Survey Data

Estimative Words of Probability Trends in National Intelligence Estimates


 
All Academic, Inc. is your premier source for research and conference management. Visit our website, www.allacademic.com, to see how we can help you today.