3.4.1.1. Overall RTB.
The hypothesis that people who are anxious are risk averse received limited support. Results in Table 19 show that anxiety level has a significant negative correlation with everyday risks (total score of the ERI-E), but only for the community sample. Given the correlations between socio- demographic variables and everyday RTB, multiple regression analyses were undertaken, which showed that anxiety level was not a significant predictor of everyday risks once the effects of these other variables was taken into account. This was primarily due to people on low incomes being both risk averse and anxious. Income is a better predictor of RTB than anxiety level, with everyday RTB having a higher correlation with income (.216) than anxiety level (-.164). Once the affects of income were taken into account anxiety level did not explain a significant amount of the variance in the overall everyday RTB score. In other words, knowing a person’s income level enables us to predict their level of RTB more than knowing their level of anxiety does. Reasons for the importance of income on RTB are discussed in section 3.4.3. This advances research in investigating the impact of other variables on the anxiety – RTB relationship. To investigate the possibility that the relationship between anxiety and everyday risks was not linear an ANCOVA with age as a covariate was undertaken, but this showed that everyday risk scores did not differ based on the level of anxiety (Section 3.3.3.1). In summary, although anxiety level does have a significant relationship with everyday risks this was for only one sample (community sample), and it is not an important predictor of risk aversion for everyday
risks when other factors are taken into account. Given these results the hypothesis from the research of Eisenberg et al. (1998), and Maner et al. (1997) that anxious people are risk averse is only partially supported.
Further analysis of previous research indicates that the observed correlations between RTB and measures of anxiety are typically 0.2 for general population samples (Maner & Schmidt, 1997, Hockey et al, 2000), similar to results for the community sample in the present research. This means that large sample sizes are required to achieve statistical significance. In contrast using clinical samples of OCD patients, even small samples showed significant differences between the groups (Rees et al., 2005; Steketee & Frost, 1994). These results of this study indicate that anxiety is a weak predictor of risk aversion, and small differences in research factors are likely to lead to different conclusions about whether the relationship is significant. These differences could include different participants, sample sizes, and the methods of measuring both RTB and anxiety. For this research the correlations were in the expected direction for each sample (anxious people were more risk averse), but this only reached the level of statistical significance for the community sample (Table 19). There is therefore some support for anxiety being a weak predictor of everyday RTB.
The hypothesis that anxiety is a stronger predictor of risk aversion than depression was partially supported. Results in Table 19 show that anxiety level has a higher correlation with everyday risks than depression level for all samples, showing a consistent pattern even for results which were not statistically significant. These results support previous research than has investigated these differences (e.g. Hockey et al, 2000), however the weakness of the relationship between anxiety and RTB means this conclusion should be treated with some caution.
3.4.1.2. Domain specific RTB.
Results show that there is some domain specificity in the relationship between RTB and anxiety, correlations in Table 19 indicate that some domains have stronger relationships with anxiety than other domains. The domains with the strongest relationship with anxiety are risks to belongings, health risks, and risks involving personal danger. This partially supports the hypotheses that predicted anxious people would be more risk averse in some domains of RTB. There is support for the hypothesis of risk aversion for health risks (Butler & Mathews, 1983), but not for the hypothesis of risk aversion for unknown risks, developed from the research investigating intolerance of uncertainty (IU) (Dugas et al, 1997). It does support the hypothesis based on the results of Stober (1997) which predicts anxious people being less likely to avoid risks to others. These results advance research in this area by specifically investigating domain specific RTB suggested by other research, and finding some support for the relationships this research indicated should be present.
Multiple regression equations were used to determine which variables explained a significant amount of the unique variance in everyday RTB domains. Results of this analysis generally supported the findings of the correlational analyses in regards to the importance of anxiety level in some domains. For two of the three domains that had significant correlations for anxiety level (for the community sample), anxiety level also explained a significant amount of the variability in scores once the effects of socio-demographic variables was taken into account. For both risks involving personal danger and health risks anxiety level plays a significant role in predicting people’s RTB. Furthermore, although there is some variability in the types of risks avoided by anxious people, all correlations were negative, and of similar magnitude, so this does not in general support the hypotheses of Butler and Mathews (1983) that anxious people will show more variation in the types of risk they will avoid.
Analysis dichotomising the participants on the basis of age did show that the relationship between anxiety and everyday RTB was stronger for older people than younger people. In people aged over 40 anxiety level explained a significant amount of the variability in overall everyday risks score, as well as for three of the six domains. In particular, for health risks anxiety level was the most important explanatory variable, accounting for 7.2% of the unique variability in scores, anxious people were more risk averse. For the younger age group the only variable for which anxiety level explained a significant amount in variability was 2% for social risks. Anxiety level is therefore a significantly more important predictor of risk taking in older adults (over the age of 40) than younger adults in this study. Given the domain in which anxiety has the largest impact is health risks, it may be that people with health concerns are both
anxious, and avoidant of risks to their health. It may also simply reflect the presence of items on the anxiety scale of the DASS-21 measuring physical symptoms, which in this population are simply results of their health conditions rather than symptoms of anxiety.
The general conclusion in regards to anxiety and everyday risk taking is that although anxious people are more risk aversive, the relationships are weak, and only occur for older adults (aged over 40). Socio-demographic factors play a significantly more important role, and anxiety has a weak association with everyday risk taking scores once the effect of these factors is taken into account. This is the case for overall everyday risks, as well as for each domain. There is some variability in the domains of everyday risk taking that anxious people avoid (see Table 19), but it is likely that this is more due to chance variability than any population differences. This is because no domain has a consistent pattern of significant scores for different samples, and comparing the correlations of each of the domains with anxiety level, the results are relatively consistent for each sample. The main exception to this general conclusion is that older people
who are anxious are significantly more likely to avoid health risks (see previous paragraph for possible explanation for this result).
This research advances knowledge in this area in specifying the domain specific relationships between anxiety and everyday RTB. It also identifies specific features of this relationship for older adults that do not exist for younger adults, especially in the area of health risks.