More Happy or Less Unhappy? Comparison of the Balanced and Unbalanced Designs for the Response Scale of General Happiness

Abstract

The single-item measure of general happiness has been widely used in questionnaires due to the advantages of easy implementation in surveys for comparison across time and culture. The balanced response scale that includes equal positive and negative response categories based on Likert-type response format has been commonly applied. However, the possibility of using an unbalanced response scale for happiness, for instance, more choices on the happy side, has not been fully examined. This study aims to explore the optimal number of response categories and the corresponding labels for general happiness by using telephone survey data in Taiwan. Six types of response scales with different combinations of response number and response labels were examined to distinguish both the intensity and direction of responses. A completely randomized experimental design using computer-assisted telephone interviewing system was conducted to collect data from representative samples aged 18 years or older. Individual characteristics among the six groups indicated that all of the sub-samples were similar in terms of gender, age, education, marital status, working status, and monthly income. Results of the graded response model suggested that a scale with at least three response categories on the positive side and no more than two on the negative side will be suitable for the single measure of general happiness. Findings of ordered logit regression on happiness were also in favor of an unbalanced response design. A discussion of the result is provided.