The Tripartite Model of Risk Perception (TRIRISK): Distinguishing Deliberative, Affective, and Experiential Components of Perceived Risk

The reliabilities and ranges for the TRIRISK scales were similar for the cancer vs. no cancer groups. Among participants with no diagnosis of cancer, the standardized (z-score) deliberative scale (α = 0.917) has a range of −2.26–2.38, and skewness of −0.054 (SE = 0.193);, the standardized (z-score) affective scale (α = 0.972) has a range of −1.88–1.51, and skewness of −0.080 (SE = 0.193); and, the standardized (z-score) experiential scale (α = 0.873) has a range of −2.77–1.62, and skewness of −0.353 (SE = 0.193). Among those with cancer, the standardized (z-score) deliberative scale (α = 0.832) has a range of −2.23–2.29, and skewness of −0.248 (SE = 0.203); the standardized (z-score) affective scale (α = 0.967) has a range of −1.88–1.51, and skewness of −0.469 (SE = 0.203); and the standardized (z-score) experiential scale (α = 0.841) has a range of −2.77–1.62, and skewness of −0.716 (SE = 0.203). To test the scale distributions for the cancer vs. no cancer groups, we divided skewness values by their standard errors. Only two values exceeded the threshold of 1.96, namely, the affective and experiential scales for participants who had been diagnosed with cancer. We therefore transformed these variables which reduced skewness to −0.266 and −0.216, SE = 0.203 and 0.203, respectively [52]. We then recomputed the correlations; r values were virtually unchanged.

We also tested whether the three risk components statistically mediated the impact of cancer diagnosis (yes, no) on intention using the Preacher and Hayes SPSS Macro for Multiple Mediation with 5000 as the number of bootstrap resamples [53]. Diagnosis reliably predicted deliberative, affective, and experiential risk perceptions (B = 0.44, 0.42, and 0.51, SE = 0.11, 0.11. and 0.11, ps < 0.001) and deliberative, affective, and experiential risk perceptions reliably predicted intention (B = −0.76, 0.48, and 0.34, SE = 0.09, 0.10, and 0.11, ps < 0.001, < 0.001, and 0.003). The direct effect of diagnosis on intention remained significant controlling for risk perceptions (B = 0.36, SE = 0.14, p = 0.013). Importantly, however, the indirect effects of deliberative, affective, and experiential risk perceptions each proved reliable (estimates = −0.34, 0.20, and 0.17; 95 % CI = −0.55 to −0.17, 0.09 to 0.36, and 0.06 to 0.35, respectively). These findings are consistent with the hypothesis that the three components of the TRIRISK model mediate the impact of cancer diagnosis on cancer-protective intentions.

Because (a) some of the deliberative risk perceptions could be seen as conditional on current behavior (e.g., “The way I look after my health means that my odds of getting cancer in the future are . . .”), and (b) conditional risk perceptions can be a strong predictor of future behavior, we examined correlations between each individual deliberative risk perception item and intentions in the full sample to see if items performed similarly. Items 3, 4, and 5 each had a statistically significant negative correlation with intentions (rs = −0.254, −0.183, −0.264, respectively; all ps < 0.001). However, item 2 also had a marginally significant, negative association with intentions (r = −0.108, p = 0.062); the other items had non-significant correlations with intentions that were mostly negative. Thus, while conditional items have some role in explaining the negative correlation between deliberative risk perception and intention, this does not entirely explain the negative correlation between deliberative risk and intention.

We also tested the interactions between cancer diagnosis and risk components but none of the interaction terms proved reliable. Deliberative, affective, and experiential risk perceptions were equally powerful and reliable predictors of intention for participants with and without a diagnosis of cancer.