Visualizing Item Nonresponse for Large-scale Surveys and High-stakes Examinations (SEER Method)

• Activity Number: 218
• Type: Contributed
• Date/Time: Tuesday, August 10, 2004 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistical Graphics
• Author(s): Chris Chiu*+
• Companies: Law School Admission Council
• Keywords: data visualization ; item nonresponse ; matrix algebra & spare matrix ; statistical computing ; Law School Admission Test (LSAT) ; national survey

Abstract

Item nonresponse is a threat to large-scale survey. Results derived from survey questions with high item nonresponse rates can lead to biased and misleading conclusions about estimates of national demographic, housing, social, and economic characteristics as well as the labor force in the country. Despite the substantial efforts and resources used to reduce item nonresponse, such as the use of Computer Assisted Telephone Interviews (CATI), Computer Assisted Personal Interviews (CAPI), item-skipping prevention methods (e.g., cognitive methodology, use of good information and organization, use of respondent-expected-skipping sequence), it is impossible to totally avoid item nonrespnse in large-scale surveys because item nonresponse is an artifact that can be attributed to many factors. Some of the factors cannot be completely controlled by the survey administers (e.g., respondents refuse to respond, respondents do not know the answers, respondents incorrectly skip an item). The proposed research project sets out to develop a set of statistical and computing tools with an ultimate goal of empowering the Bureau and other agencies to become more proactive in this issue.