A partial ordering application in synthetizing dimensions of subjective well-being

The present work originates from a study on the relationships between labour status (LS) and subjective well-being (SWB). It analy, ses Italian data from the European Union Statistics on Income and Living Conditions (EU-SILC) 2013. In that edition, Eurostat adopted an ad-hoc module on SWB, inspired by the Guidelines on Measuring Subjective Well-being (OECD, 2013). These guidelines recommended to consider the three main dimensions of SWB: cognitive, affective and eudaimonic. According to OECD’s Guidelines, the three main dimensions of SWB refer to non-elementary concepts. The Guidelines suggest choosing indicators able to represent the multidimensionality of the concepts. The patterns of analysis must respect this multidimensionality, and the choice of data processing methods must conserve the informative potential of each item (Maggino, 2015). Eurostat (2015) formulated its analysis comparing national aggregate data. We chose to analyse the micro-data. We applied the analysis to quite 16 thousands of respondents. In order to represent the three dimensions of SWB, we chose: 1) the question on satisfaction with life which is representative of the cognitive dimension; 2) the question on meaning of life, which is considered by Eurostat as a proxi of the eudaimonic dimension; 3) the five items on emotions occurred to respondents during the previous four weeks, which represent the affective dimension. We started with processing the five ordinal variables describing emotions. We saw that the correlation between the five items concerning affects, were not so high. This result and other observations on conjoint distribution of levels of emotions, showed that there were a significant number of respondents which are both happy and sad, serene and nervous. We decided to apply a Partial Order methodology to aggregate the five dimensions of emotional status. This methodology allows, in fact, to deal with ordinal variables concerning multidimensional phenomena and to analyse complex relationships at micro level. We defined a Partially Ordered Set (POSET) using Parsec, which is a package developed in R by A. Arcagni and M. Fattore (2014). In that way we obtained a synthetic dimension which we named Emotional Status. Eventually, applying the Partial ordering methodology to the three dimensions of subjective well-being, we calculated a subjective well-being synthetic index. The last step of this analysis compare the results of this method with those of other synthesis methods.