Design
A cross-sectional internet survey study was conducted.
Participants
The participants of the survey were general residents recruited through the survey system offered by Macromill (https://www.macromill.com/). The candidates were invited to the survey if they were general residents (no particular restrictions on health conditions, such as past illnesses), in the age group of 20 to 65 years, and were able to understand the purpose of the survey and accordingly provide their consent. There were no particular exclusion criteria.
Procedures
We conducted the internet-survey using the system provided by Macromill. They sent an email to the registered samples to announce this study. The candidates who showed interest and were ascertained as eligible for the study were offered a written leaflet which provided detailed information about the research. Once they provided their consent through the survey system, they were requested to fill four self-reported questionnaires related to well-being.
Clinical scales
The scales included in the survey were ICECAP-A, SWLS, FS, and SPANE. It is to be noted that the notion of well-being is complex and there is no consensus among researchers regarding its parameters. However, to date, it is thought that well-being consists of different aspects and measuring such aspects is crucial in providing comprehensive snapshots of the respondents’ well-being (OECD, 2013).11 The Organisation for Economic Co-operation and Development (OECD) has prescribed guidelines on measuring subjective well-being. Therefore, in this survey, we requested participants to fill the following three clinical scales, in addition to ICECAP-A, each of which measures a different aspect of well-being (SWLS: cognitive; SPANE: affect; FS: eudemonic aspect of well-being).
ICECAP-A: ICECAP-A was developed to measure the capability well-being of adults, which was not captured adequately in pre-existing health-related quality of life scales. It consists of 5 attributes: attachment (ability to have love, friendship, and support), stability (ability to feel settled and secure), achievement (ability to achieve and progress in life), enjoyment (ability to experience enjoyment and pleasure), and autonomy (ability to be independent), each of which has four levels. It provides a single index value for well-being utility ranging between 0 and 1. A higher score indicates a better well-being status.8
SWLS
This scale is a 5-item self-reported questionnaire to evaluate the cognitive aspect of subjective well-being. Scores for each subscale range from 1 = strongly disagree to 7 = strongly agree. Total scores range from 5 to 35, with higher scores indicating higher satisfaction.12
FS
This scale includes eight items relevant to significant aspects of human functioning, ranging from positive relationships to feelings of competence, meaning, and having a purpose in life. FS is recognized as a scale which measures the eudemonic aspect of subjective well-being. Each item is answered on a 1–7 scale that ranges from strong disagreement to strong agreement. Possible scores range between 8 and 56. A higher score indicates that the respondent has a better eudemonic well-being.13
SPANE
SPANE consists of twelve items, including six items to assess positive experiences and six items to assess negative experiences.13 The positive score (SPANE-P) and the negative score (SPANE-N) are in the range of 6 to 30. A higher score means higher positive or negative affective aspects of well-being. The two scores can be combined by subtracting the negative score from the positive one (SPANE-B, score range: -24-24).
Statistics
In statistics, to predict the score of ICECAP-A, a beta regression was conducted, with utility assessed by ICECAP-A as a dependent variable, because this score ranges from 0 to 1, which is beta distribution. Since the tariff of ICECAP-A among Japan’s population is not still available, we used the weighted utility scores by ICECAP-A among UK samples. Age, sex, SWLS, FS, and SPANE were set as independent variables. Several statistical models were developed (i.e. A-G: total score was used, a-g: each sub-score was used) (Table 1), and fitting levels were compared using the mean absolute error (MAE) and the root mean squared error (RMSE) because those error values provide better means to assess mapping functions than R-squared, which focuses on how well the model explains the dataset it was estimated on.14 The independent variables in each model were as below:
<Independent variables in each model>
Model name:
-Uppercase means total score was used, while lowercase means each subscale was used
-Number means the number of questionnaires used in the model
- model 1A/a: SWLS, model 1B/b: FS, model 1C/c: SPANE-P, model 1D/d: SPANE-N
- model 2A/a: SWLS and FS, model 2B/b: SWLS and SPANE-P, model 2C/c: SWLS and SPANE-N, model 2D/d: FS and SPANE-P, model 2E/e: FS and SPANE-N, model 2F/f: SPANE-P and SPANE-N
- model 3A/a: SWLS, FS and SPANE-P, model 3B/b: SWLS, FS and SPANE-N, model 3C/c: SWLS, SPANE-P and SPANE-N, model 3D/d: FS, SPANE-P and SPANE-N
- model 4A/a: SWLS, FS, SPANE-P and SPANE-N
All the participants were divided into two groups. Three hundred participants were randomly selected, and their data were used for the development of the model (estimation sample). The data of the remaining participants were used to assess the validity of the developed model (validation sample). These analyses were conducted using R (4.1.0).