Chapter 4: Predictors of well-being

The advantage survey data holds over other, aggregated forms, is that we can go really bore down and unpick patterns of well-being based on individual living conditions and variables. Furthermore, using regression techniques, we can analyse the impact of several variables at the same time, allowing a full picture. For example, whilst analyses in Chapter 3 might suggest that income has a positive impact on a certain aspect of well-being, we might find that this effect does not emerge if we control for, let’s say, education levels.

Exercise 12

Let’s try an example. Perform a multiple regression with vitality as the dependent variable, and the following independent variables: gender, income (use the 12-band variable), level of education, and time spent watching TV. Remember, as usual, to weight correctly (in this case you need the combined weight as you are pooling across countries).

Solution SPSS

Using SPSS, you should get the following results:

Table 4.1. Coefficients
B* Std. Error* Beta** t Sig.
(Constant) .036 .020 1.784 .074
TV watching -.024 .002 -.082 -12.187 .000
Gender -.131 .008 -.112 -17.016 .000
Income .029 .002 .122 17.265 .000
Education .029 .003 .076 10.563 .000
  • Dependent variable: Vitality
  • *Unstandardised Coefficients
  • **Standardised Coefficients

All four variables are strongly significant. This shows that higher incomes and levels of education are associated with higher levels of vitality, whilst watching more TV is associated with lower levels of vitality. To interpret the gender variable you need to check how it is coded. In this case ‘1’ indicates male and ‘2’ indicates female. This means that, on average, females have vitality scores 0.131 lower than males. To generate these results, we used the following independent variables: TVTOT, GNDR, HINCTNT and EDULVL.

SPSS syntax

/DEPENDENT vitality
/METHOD=ENTER tvtot gndr hinctnt edulvl.
Go to next page >>