Back in September
a PSC Research Project
Investigator: Frederick G. Conrad
Little research has evaluated the role of speech cues and vocal qualities of respondents and interviewers as they pertain to data quality in survey research. No research to date has used speech and voice to predict item-level data quality (e.g., item nonresponse and response inaccuracy). With advances in the sciences of cognitive and social psychology and the scientific study of speech, we can now develop more advanced social psychological models of data quality. The true cause of item nonresponse and measurement accuracy remains elusive (Census Dissertation Fellowship RFP Topic J), and the use of speech and voice variables to explore the interviewer-respondent (I-R) interactions that lead to item nonresponse and measurement inaccuracy may lead to new insights (Census Dissertation Fellowship RFP Topic G). Time series models predicting changes in speech and voice qualities over the course of individual interviews will be used to predict item nonresponse and accuracy outcomes for target sensitive questions. Particularly, I am interested in whether speech cues and vocal behaviors that occur before the target question is read or those that occur after the question is read will have more impact on item nonresponse and inaccuracy. This research has implications for psychological models of survey response as well as interviewer training. Frederick Conrad, Research Associate Professor, Institute for Social Research, University of Michigan will serve as advisor and chair.
|Funding:||U.S. Bureau of the Census|
Funding Period: 06/04/2008 to 06/03/2010
Country of Focus: USA