Optimally Scaled Quality-of-Life Indicators

When studying quality of life, researchers have to rely on the subjective evaluations, which are typically categorized, collected in surveys. When statistical analysis are applied to these data, they used to apply the simplistic approach including (i) direct quantification, which assigns discrete numerical values to ordinal response scale, and (ii) complete-case analysis, which discards all observations selecting any of the off-scales choices like Dont Know No Answer from the analysis. The present paper examines the disadvantages of this approach and introduces the optimal scaling method as a remedy. The new scheme attempts to restore the continuity property of the measurements as well as provide estimates for most of the missing responses. Application of the new scheme to the Hong Kong QOL data illustrates how the scheme works, demonstrates its advantages and shows how the QOL indicators, the global QOL indicator as well as its inherited indicators, can be constructed from series of principal component analyses. Factor analysis of the 20 life domain indicators verifies Wan's opinion (1992) that the global sense of well-being can best be captured by two dimensions, namely the personal well-being and the societal well-being. Although previous QOL studies had seldom included perceptions towards various societal conditions to identify life satisfaction in general, our data analysis shows that satisfaction on these conditions do constitute an important component of the global QOL.