Invalid Answers

TANAKA Sigeto (tsigeto(AT)nik.sal.tohoku.ac.jp)

SIMIZU Shinji (ed.), 2001, Kazoku seikatu nituiteno zenkoku tyousa (NFR98) No.2-4: Gendai Nihon no kazoku isiki [NFR98 Report No. 2-4: Family consciousness in the contemporary Japan]. NFR Kenkyuukai, Japan Society of Family Sociology (c/o OKUBO Takaji, Faculty of Letters, Waseda University). pp. 155-179.

Grants-in-Aid for Scientic Research results report submitted to the Ministry of Education, Science, Sports and Culture: No. 10301010.

National Institute of Informatics online cataloging system (NACSIS-CAT) number (NCID): BA53222707

Abstract

In this paper, the term "invalid answer" (IA) means such choises as "sono ta" (NAC: there is no appropriate choice) and "wakaranai" (DK: don't know), as well as "no answer". From the NFR98 questionnaire, 50 questions are subjected to scrutiny because they are applied for all of the sample. Findings are as follows:

3,214 cases (46% of the valid sample) have IA.
There are some districts where both valid reception and IA show high rate.
Those who have such characteristics as follows show high IA rate:
- aged people with preliminary or secondery education;
- those who have no job;
- unmarried people.
Questions offering NAC/DK choices tend to create IA.
Canvasser's checks at the picking up the questionnaire reduce the probability of IA occurance.
Same kind questions consist a sequence show correlations in IA occurance.
Education have the effect of reducing IA on the all 50 questions.

Based on the findings, this paper mentions to caution in analizing NFR98 data, and to modifications to similar surveys in future.

Key Words: missing data, no answer, non-sampling error

[Body of the paper (Japanese; 251KB PDF)]
This PDF file is identical to the published paper, though it has no pagenation.