A New Measure of Life Satisfaction: The Riverside Life Satisfaction Scale
Seth Margolis, Daniel Ozer, Sonja Lyubomirsky, and I have designed a new measure of overall life satisfaction. We believe that this measure improves on the most widely used multi-item measure of life satisfaction, Diener et al.'s Satisfaction With Life Scale (SWLS) (Diener, Emmons, Larsen, & Griffin, 1985).
Diener's SLWS consists of the following five questions, each answered on a 1-7 scale from "strongly disagree" to "strongly agree":
SWLS items:
In most ways my life is close to my ideal.
The conditions of my life are excellent.
I am satisfied with my life.
So far I have gotten the important things I want in life.
If I could live my life over I would change almost nothing.
One often-noted feature of the SWLS is that the first four items are direct measures of current life satisfaction, whereas the fifth item concerns regret about the past. Accordingly, the fifth item generally has lower factor loadings onto the scale than the first four items, which are a more tightly clustered group that people tend to answer similarly. A more unified construct might cut the fifth item.
However, we like the fifth item. Instead of cutting it, we think a good measure of life satisfaction should have more items like it, creating a broader target. To see why, consider a hypothetical respondent who answers the first four SWLS questions with "strongly agree" but who also feels intense envy of others' lives, is full of regrets, and wants to change their life path. Such a respondent, we think, should not be regarded as having maximum life satisfaction. Envy, regret, and desire to change are all either indirect signs of, or direct constituents of, dissatisfaction. A good measure of life satisfaction should include questions about them.
It is this issue -- how should we conceptualize (and thus measure) life satisfaction -- that's my own central concern in this project. Readers familiar with my general approach to attitudes will be unsurprised to hear that I favor a broad, dispositional approach to overall life satisfaction. A person truly satisfied with their life is not just someone who is disposed to sincerely say "I'm satisfied" but someone who is also disposed to act and react generally in ways concordant with being satisfied, i.e., not desperately seeking to change their circumstances, not seething with envy at others' situations, etc.
Also, in constructing a new measure, we wanted to have some reverse-scored, negatively-valenced items to mitigate acquiescence bias (that is, a tendency to say "yes" to most questions).
Thus, we created a scale with three SWLS-like items and three reverse-scored, negatively-valenced items targeting envy, regret, and desire to change. The items are presented in random order and answered (like the SWLS) on a 1-7 scale from "strongly disagree" to "strongly agree".
RLSS items:
I like how my life is going.
If I could live my life over, I would change many things.
I am content with my life.
Those around me seem to be living better lives than my own.
I am satisfied with where I am in life right now.
I want to change the path my life is on.
To test this new measure, we ran three studies (all described in our final manuscript version). In Study 1, we selected the final six items above from a larger pool. Across the three studies, we examined inter-item correlations and factor loadings among our six selected items, and we correlated our new measure with the SWLS, the Big 5 personality traits, several demographic variables, measures of positive and negative emotions or affect, the Psychological Well-Being Scale, the Schwartz Values Survey, and a measure of socially desirable responding.
The details are available in our paper, but a few highlights are:
(1.) Despite our use of the three negatively-valenced indirect items targeting a broader underlying phenomenon, the SWLS and RLSS showed high and almost identical levels of internal consistency, suggesting that our addition of these items didn't harm the coherence and unity of the measure.
(2.) The RLSS correlated highly with the SWLS (r = .88, .89), suggesting that most existing research relying on the SWLS would generate similar results had the RLSS been used instead. Thus, despite (we think) improvements, it is not targeting a radically different phenomenon.
(3.) Finally, the RLSS consistently correlated slightly better than did the SWLS with measures of self-reported emotions, the Psychological Well Being Scale, and the Big 5 personality trait of Negative Emotionality. Thus, arguably, the RLSS is slightly more centrally located than the SWLS in this network of related psychological phenomena.
[Thanks to Dan Haybron's Templeton grant for funding!]