As depicted in Figure 2, age was inversely associated with preferred choice set size (Spearman’s ⫽ ⫺ .29, p ⬍ .01). The negative correlation between age and choice preferences was significant for all domains except for jam varieties ( ⫽ ⫺ .03, ns ). Post hoc t tests indicated that middle-age participants (ages 40 –59 years; n ⫽ 74) desired significantly more choices among varieties of jam ( M ⫽ 10.6, SD ⫽ 8.2) than older (ages 60 ⫹ years; n ⫽ 108) 1 Correlation coefficients for pairs of items in the Big Five Inventory ranged from r ⫽ .24 (Openness) to r ⫽ .53 (Extraversion), p s ⬍ .01. 2 Attribute information was retrieved from Edmunds.com, the National Highway Traffic Safety Administration Website, and the JD Power and Associates Website. We deliberately chose to omit pricing information from the grid so that participants did not simply select the most expensive car. For some cars, it was not possible to obtain full information for all six attributes in those cases, the missing attribute was calculated as the mean attribute value for all other cars in the information grid. 3
At this time, all participants who completed the behavioral task were randomly assigned to receive instructional manipulations designed to raise or lower their decision-making self-efficacy levels (for details on the specific instructions, see Reed et al., 2012). However, because these manipulations did not significantly affect DMSE, choice preferences, or information search, further analyses collapsed the sample across experimental conditions.
4 We opted against adjusting the size of the information grid to participants choice set size preferences to avoid confounding measures of choice set size preferences and information search. 4 REED, MIKELS, AND LÖCKENHOFF
participants ( M ⫽ 8.3, SD ⫽ 6.6) but not younger (ages 18 –39 years; n ⫽ 133) participants ( M ⫽ 9.8, SD ⫽ 8.3). For all other domains, correlations between age and choice set size preferences ranged in size from (vacations) to (physicians; p s ⬍ .05) and the size of this association did not differ systematically between everyday ( ⫽ ⫺ .26), and health-related domains ( ⫽ ⫺ .25). Consistent with our hypothesis of a linear age trend, regression analyses indicated that neither the quadratic nor cubic effects of age on choice preferences were significant. Thus, the association between age and choice set size preferences appears to be linear and generalizable across domains.
Consistent with previous research, age was negatively associated with maximizing, memory self-efficacy, future time perspective, neuroticism, STM (digit span, and numeracy, but positively associated with conscientiousness and vocabulary (see Table 3). In addition, age was positively associated with DMSE and preference accessibility, but not significantly associated with need for cognition or optimal choice beliefs.
To examine the role of covariates, we computed separate regressions with choice preferences as the dependent variable and age and each of the covariates as the predictors. As depicted in Table 3, age remained a significant predictor when each of the covariates was statistically controlled ( p s ⬍ .01), suggesting that none of them could account forage differences in choice preferences. Of all the covari- ates, only vocabulary ( ␤ ⫽ ⫺ .14) and optimal choice belief ( ␤ were significantly associated with choice preferences when controlling forage, indicating that, regardless of age, individuals with lower vocabulary scores and greater confidence in the benefits of choice preferred larger choice sets.