Computing overall satisfaction
There is discussion in the literature about whether measures of overall satisfaction should be calculated as some form of weighted composite of facet satisfaction where weights are derived from importance ratings.
These debates have been applied to various types of satisfaction including life, job, and customer. In this context, it does not seem like weighting by importance makes much difference, and it may even be detrimental.
For examples, Trauer et al (2001) state that:
Many Quality of Life (QoL) instruments ask respondents to rate a
number of life domains in terms of satisfaction and personal
importance, and derive weighted satisfaction scores by multiplying the
two rat- ings. This paper demonstrates that this practice is both
undesirable and unnecessary. QoL domains are selected on the basis of
their inherent importance, rendering separate importance rating
partially redun- dant. Weighted scores present difficulties in
interpretation. Further, we show that multiplicative composites have
undesirable psychometric properties. There is evidence that
multiplicative composites have little or no advantage over unweighted
ratings in correlational or predictive studies. Apart from the face
validity and the intuitive appeal of multiplying satisfaction ratings
by importance ratings, there appear to be no sound reasons for doing
so, and several good reasons not to do so.
A similar critique of importance ratings in the job satisfaction domain can be seen in Blood (1971).
Nonetheless, this all assumes that importance measurement is a means for forming an overall composite. In some instances, measurement of importance is of substantive interest. Trauer et al (2001) acknowledge this when they write:
It is not suggested that the assessment of the importance of domains
of satisfaction is inconsequential. It has been shown that domains
that are closer to the personal core (e.g. health, marriage, family)
are typically judged as more im- portant than more peripheral domains
(e.g. membership of organizations)...
Study of mean importance ratings
Importance ratings can be used to assess average perceptions of importance which might inform causal hypotheses regarding what is influencing satisfaction.
Take for example, Rice et al (1991) among many other things, they report importance ratings of various job facets. Among other things the authors asked 97 employed college students to rate
facet importance on a 9-point scale with two verbal
anchors: not at all important and extremely important (e.g., "How
important to you is the amount of opportunity for promotion?").
If we sort the results reported in Table 1 by importance we see the following:
Job Facet Mean Importance
Hourly pay (dollars) 7.4
Hours per week 7.3
Control over schedule 7.3
Opportunity to learn skills 6.7
Decision making 6.4
Promotion opportunity 6.3
Mental effort required 6.1
Conversation with co-workers 6.0
Customer or client contact 5.7
Physical effort required 5.4
Supervisor contact 5.3
Commuting time (minutes) 4.3
This is presumably an interesting empirical finding in itself.
Relationship between importance and satisfaction
Some researchers have explicitly studied the relationship between importance and satisfaction to see whether there is a relationship. E.g., Borg (1991) provides an example of job satisfaction exploring how intra-individual variability in facet job satisfaction relates to facet importance. They find support both for the idea that more important facets are deemed more satisfying and for a u-shaped relationship where it is the most and least satisfied facets that are the most important. This variation depends on the domain being studied.
- Blood, M.R. (1971). The validity of importance.. Journal of Applied
Psychology, 55, 487.
- Borg, I. (1991). On the relationship between importance and satisfaction ratings on job aspects. Applied Psychology, 40, 81-92.
- Rice, R.W., Gentile, D.A. & McFarlin, D.B. (1991). Facet importance and job satisfaction.. Journal of Applied Psychology, 76, 31.
- Trauer, T. & Mackinnon, A. (2001). Why are we weighting? The role of importance ratings in quality of life measurement. Quality of life research, 10, 579-585.