(From the article of A. Andreenkova, W. Saris)

During the 80’s several studies tried to determine whether satisfaction with life as a whole and happiness measure the same or capture different sides or levels of subjective evaluation of life. Campbell, Converse and Rogers (1976) found that the correlation between these two measures was .5. Michalos found for the 39 countries, he studied, a correlation around .6. McKennel and Andrews (1980) suggested that satisfaction differs from happiness because the former represents the cognitive component of the subjective evaluation while the later represents the emotional side of it. However, Schwarz and Clore (1983) could not reproduce the last result. Moreover Diener (1984) suggested that the difference between these two indicators can not be due to differences in the cognitive or emotional nature of each of the indicators only. He suggested that the relationship is more complicated.

We think that the relationship between these two indicators is varying with the cultural and linguistic environment in which it is studied. The differences between the indicators will depend very much on the way these words are used in different languages and different countries. Therefore we suggest to study the relationship between these two variables in Russia and to draw conclusions for this country and language without a suggestion of a possible generalisation.

During the last two waves of the RUSSET panel we have asked satisfaction as well as happiness questions twice in each wave. Having repeated observations, we can correct the responses for measurement error and test if these two questions indeed measure the same concept. If that would be the case the correlation between the variables corrected for measurement error should be 1. If the correlation is significantly lower than 1, they measure different things (Jöreskog 1971). Applying this approach to satisfaction and happiness questions, we found that the correlation between them is indeed far from 1, namely .64. This finding suggests that the overlap between the responses to satisfaction and happiness questions in Russia is rather low (even after correction for measurement error it is only around 40%).  So these variables really don’t measure the same concept in Russia. 

Other evidence for the fact that these two variables are really measuring something different is that the difference in chi2 test, comparing models with and without the restriction that the correlation between the variables is 1, is highly significant with a chi2 value of 148 for 1 degree of freedom. This result clearly leads to the rejection of the hypothesis that these variables measure the same concept. 

This conclusion leads to another question i.e.: In what respects do these two questions differ? In order to answer this question we looked at the correlation between satisfaction and happiness questions with other variables, which indicate feelings or cognitions, trying to find the pattern mentioned above. The variables used in this test are described in table 3. One could expect that the first 7 variables, which measure feelings, would correlate more strongly with happiness, while the satisfaction variables, being cognitions, would correlate higher with SWB.

First we tested a model with the assumption that all correlations were equal. This model had to be rejected. Then we looked at the ‘modification indices’ and ‘expected change parameters’ to determine which correlations caused the most serious problems. Two variables: ‘satisfaction with marriage’ and ‘satisfaction with the financial situation of the household’ were the variables that could improve the fit the most, when they were allowed to be different for the two variables. Leaving these two correlations free to be different for satisfaction and for happiness, a difference in chi2 of 219 for 2 degrees of freedom was obtained. After that change in the model, no further serious improvements could be made even though the sample size (3400) was quite large. These tests did not confirm the expectations: the correlations between the two sets of  variables and the responses to the satisfaction and the happiness questions were equally strong. The only difference was that satisfaction with marriage correlated more strongly with happiness and satisfaction with the financial situation with SWB. Because feeling questions do not differentiate between SWB and happiness, the hypothesis that happiness measures mainly a feeling and satisfaction a cognition is not confirmed. This analysis shows that Russians associate happiness more with the success of their marriage (internal factors, social and personal factors) while satisfaction with life as a whole (SWB) is more strongly connected with the financial situation (external factors, material factors) but otherwise these two variables are very similar in their relationship with the specified feelings and cognitions.

Table 3 The correlations between happiness and satisfaction and other variables