주거환경만족도 평가의 신뢰성 향상을 위한 기초연구 - 주거실태조사의 설문구성 방식을 중심으로 -

Abstract

This paper aims to investigate the reliability verification of residential satisfaction evaluation in terms of the SP (Same-Pattern) responses suspected as outliers under the 2006-2017 KHS. Based on the research hypothesis developed from the distributional characteristics in the Same-Pattern (SP) responses obtained across 2006-2017 KHS, each of three distinctive questionnaires types (type B, R, and D) are prepared and distributed to 750 respondents (2,250 respondents in total) through online. The key findings are as follows: First, the "Reduced type R", in which the number of questionnaire items is reduced in an attempt to lower the likelihood of unreliable answers resulting from the fatigue effect, adversely affected the result and presented even greater tendency towards SP responses. It is concluded that such simple measures of reducing the number of questionnaires do not effectively address the problem of SP occurrence that is suspected of being outliers. Second, the "Divided type D", in which the questionnaire items were divided into multiple sub-groups in an attempt to increase the awareness of respondents and prod them into action, presented the lowest occurrence of SP responses. Moreover, this type is likely to have the lowest possibility of co-existing outliers, the result confirmed by the verification process using statistical and machine-learning analysis. Third, it became clear that reducing the number of questionnaires while simultaneously inserting heterogeneous intermediary questionnaires is the most effective method to conduct the table-formulated satisfaction evaluation utilizing a Likert scale. However, since an excessive partition may cause unnecessary psychological pressure on respondents and increase the likelihood of SP outliers, it is highly recommended to utilize the carefully designed framework for the best result.