The Multidimensional Existential Meaning Scale: A tripartite approach to measuring meaning in life

Abstract

To address conceptual difficulties and advance research on meaning in life (MIL), it may be useful to adopt a tripartite view of meaning as consisting of comprehension, purpose, and mattering. This paper discusses the development of the Multidimensional Existential Meaning Scale (MEMS), which explicitly assesses these three subconstructs. Results from three samples of undergraduates showed the MEMS to have favorable psychometric properties (e.g. good factor structure and reliability) and demonstrated that it can effectively differentiate the three subconstructs of meaning. Regression and relative importance analyses showed that each MEMS subscale carried predictive power for relevant variables and other meaning measures. Additionally, the MEMS subscales demonstrated theoretically consistent, differential associations with other variables (e.g. dogmatism, behavioral activation, and spirituality). Overall, results suggest that the MEMS may offer more conceptual precision than existing measures, and it may open new avenues of research and facilitate a more nuanced understanding of MIL.

Keywords:

• Meaning
• purpose
• measurement
• scale development
• existentialism

Notes

The present research was conducted at the University of Connecticut.

1. Note that while the Presence subscale of the MLQ is unidimensional, the full MLQ is multidimensional and contains two subscales measuring the presence and the search for MIL.

2. Due to a clerical error, demographic info was not collected for sample 3. Demographics in Sample 3 can be expected to mirror those of Samples 1 and 2 as these came from the same participant pool.

3. We do not examine if a higher order factor model – where there is a general meaning factor underlying the three subconstructs – fits the data better than a lower order model consisting of just the three subconstructs. When there are only three lower order factors, the higher order factor model would be statistically equivalent to the lower order factor model. Therefore, a comparison of the fit of the two models would not be possible.