**3. Results**

#### *3.1. Prevalence of Life Satisfaction over Time*

Life satisfaction over time (years 2006 to 2017) is illustrated in Figure 2 and full descriptive statistics can be found in Supplementary Materials (Table S2).

Between 2006–2017 mean life satisfaction varied from a high of 7.61 (*SD* = 1.6) in 2007 to a low of 7.23 (*SD* = 1.73) in 2011; the range in mean changing 0.38 over this time between the high and low.

#### *3.2. Life Satisfaction by Age and Gender*

The results of an independent samples *t*-test showed that women scored significantly higher than men on life satisfaction (*t*(10,771) = −3.763, *p* < 0.001, 95% CI of difference:

−0.186, −0.058, *d* = 0.073), albeit with a very small effect size. Figure 3 shows the distribution of life satisfaction by age and gender for all participants across all years.

**Figure 3.** Life satisfaction by age and gender.

Figure 3 shows that females tend to report being slightly more satisfied with life throughout and over the life course compared to males, with this gender gap being most predominant mid-life. Both females and males on average report slightly decreasing levels of life satisfaction from teenage to mid-life, and then increasing from mid-life onwards.

As a supplementary analysis, we also looked at gender differences in the positive and negative experience indexes we created (composite variables described above). The results of independent samples *t*-tests showed that women scored significantly higher on the negative experience index (*t*(9401.196) = −7.238, *p* < 0.001, 95% CI of difference: −0.054, −0.031, *d* = 0.146), again with a small effect size. No significant gender differences were found for the positive experience index.

#### *3.3. Other Demographic Predictors of Life Satisfaction*

Table 1 presents the results of seven separate ANOVAs, using demographic variables as independent variables (employment, education, location, religious affiliation, relationship status, country of birth, income quintile) explaining life satisfaction.


**Table 1.** ANOVA Results Predicting Life Satisfaction.

For religious affiliation, several categories with very small sample sizes (e.g., "Hinduism" and "Islam") were combined with the "other" category. As indicated in Table 1, the strongest predictor of life satisfaction was income quintile (explaining 3.7% of the variance), followed by employment status (explaining 2.9% of the variance) and relationship status (explaining 2.5% of the variance). Figure 4 presents life satisfaction and employment status.

The self-employed and those who choose to work part-time had the highest levels of life satisfaction and those unemployed experienced the lowest levels of life satisfaction. The results of the Games–Howell test showed that all employment groups were significantly different from each other (*p* < 0.05) except employed full-time for an employer and out of workforce (*p* = 0.074).

Figure 5 presents life satisfaction and relationship status.

**Figure 5.** Life satisfaction and relationship status.

Individuals identifying as married or widowed reported the highest levels of life satisfaction, and those identifying as separated reported the lowest levels of life satisfaction. In summary, and as shown in Figures 4 and 5 and Table 1, the unemployed and separated groups of individuals were in comparison the least satisfied with their lives. The results of the Games–Howell test showed that all marital groups were significantly different from each other (*p* < 0.05) except single and domestic partnership (*p* = 0.717) and domestic partnership and divorced (*p* = 0.856). A separate ANOVA indicated that gender did not moderate the relationship between relationship status and life satisfaction. The results of the Games–Howell test showed that all income groups were significantly different from each other (*p* < 0.05). Education, religious affiliation, and location each explained 0.7% or less of the variance in life satisfaction. Country of birth was not a significant predictor. The results of the Games–Howell test showed that there was a significant difference between people with elementary and tertiary education, and between people with secondary and tertiary education (*p* < 0.01). For religious affiliation, the Games–Howell test showed that all groups were significantly different from each other (*p* < 0.01), whereas for location, the significant differences were between "rural or farm" and all other types of location, as well as between "small town or village" and "Suburb of a large city" (*p* < 0.01).

## *3.4. Regression Analysis*

We included all predictors of life satisfaction along with key demographic variables (the variables are provided in the predictor column of Table 2) in a regression using the standard or simultaneous regression model. A total sample of 6023 participants, of the 10,799 participants (56%), had no missing values on all of the 28 variables and were included in the analysis (see Table 2).


**Table 2.** Comprehensive Regression Analysis.

Note. *B* = unstandardised regression coefficient. Beta = standardised regression coefficient.

The predictors collectively explained 33.1% of the variance in life satisfaction, *F*(27, 5995) = 110.024, *p* < 0.001, *R*<sup>2</sup> = 0.331. Based on the results of a separate stepwise regression analysis, household income satisfaction was the strongest predictor explaining 16.8% of the variance. The second strongest predictor was satisfaction with standards of living, contributing an additional 5.4%. Negative experience, city satisfaction, and positive experience were next explaining 3.4%, 1.9%, and 1.7%, respectively. These five variables jointly explained 29.3% of the variance in life satisfaction scores. The other variables collectively added only 3.8% of additional variance. Based on the results of the stepwise regression, perceptions of corruption, donation, religiosity, respect, education, separated, unemployed, and number of children did not add a significant amount of variance beyond the other 19 variables and were excluded.

We also conducted regression analyses separately across age and gender groups. Given the large sample size, the probability of a type 1 error is increased. Therefore, the significance threshold of 0.001 is preferred for assessing significance. Table 3 presents the unstandardised regression coefficients for age and gender groups.


**Table 3.** Unstandardised Regression Coefficients for Age and Gender Groups.

Note. HH income = satisfaction with household income; \*\*\* *p* < 0.001. Given the large sample size, the probability of a type 1 error is increased. Therefore, the significance threshold of 0.001 is preferred for assessing significance.

Satisfaction with healthcare, being separated, and unemployed were removed as they showed no variation in one or more groups. Table 4 presents the regression results across age and gender groups, and also presents the five most important predictors for each group in order of predictive power, based on separate regression analyses using the stepwise procedure for each group.


**Table 4.** Regression Results across Age and Gender Groups.

Note. The *R*2, *F*, and *df* values come from simultaneous regression analyses. The important predictors come from separate regression analyses using the stepwise method. The most important predictors are in order of predictive power. Abbreviations. SWSL = satisfaction with standards of living; HH income = satisfaction with household income; positive = positive experience; negative = negative experience. \*\*\* *p* < 0.001.

For all groups, an important predictor of life satisfaction was satisfaction with standards of living. For males across all age groups and females up to the age of 40 years, positive experiences and satisfaction with household income were also important predictors. Being married was an important predictor for males over 40 years. Feeling satisfied with city was important for females across all ages and for men under 40.

#### *3.5. Relationship between Household Income and Life Satisfaction: A Close Examination*

The relationship between per capita annual household income in International Dollars, e.g., see [44], and life satisfaction is shown in Figure 6.

**Figure 6.** The relationship between annual household income (in International Dollars) and life satisfaction in NZ.

Individuals with incomes over \$100,000 International Dollars were excluded due to their small sample size. As can be seen, the lower income group (under \$15,000 International Dollars, equivalent to about \$23,000 NZD) reported much lower life satisfaction, however, past approximately \$20,000 International Dollars (equivalent to about \$31,000 NZD), there were smaller increases as income increased.
