Saturday, November 12
Pretesting Methods
Data Quality and Measurement Error
Sat, Nov 12, 9:00 AM - 10:25 AM
Regency Ballroom-Monroe
New Approaches to Questionnaire Design and Evaluation

Conjoint Analysis Versus Split-Ballot Experiments for Survey Research (303135)

*Jennifer Childs, U.S. Census Bureau 
Casey Eggleston, U.S. Census Bureau 
Elizabeth Nichols, U.S. Census Bureau 

Keywords: conjoint, split-ballot, experimental design, administrative records, survey invitation

Conjoint measurement simulates real-world decision making by asking respondents to make tradeoffs between features in order to get the most valued combination. Respondents are presented with a series of choices between “packages” or combinations of features and asked to choose. In more traditional experimental design, a survey researcher would set up a completely crossed study in which each participant only sees one panel and results from each panel are compared. This paper examines whether it is possible to use conjoint studies in the place of costly fully-crossed experimental studies to study aspects of interest to the U.S. Census Bureau. Of particular interest to the Census Bureau is learning about factors that influence the public’s views towards the statistical use of administrative records. Additionally, the Census Bureau wants to investigate different characteristics of survey invitations that lead to increased survey responses.

The primary objective of the first study in this paper is to understand public views about the statistical use of administrative data as a substitute for survey data. The design of the study uses three attributes, each with three levels: 1) Time saved by using records; 2) what administrative records; and 3) source of the administrative records. We compare this study conducted via Choice-Based Conjoint to an experimental design.

The objective of the second study is to investigate the probable impact of different characteristics of survey notification on survey response. This study has four attributes, each with three levels: 1) Use of data; 2) description of law; 3) results of nonresponse; and 4) description of confidentiality protection. We compare this study conducted via Choice-Based Conjoint to an experimental design.