The cognitive reflection test (CRT) measures cognitive processing traits along a reflective/intuitive dimension, but it also has weaknesses as a test. A longer, integrated, non-mathematics version of this test (CRTverbal) is evaluated in terms of its ability to predict Bayesian reasoning performance and in comparison to several other individual difference measures. Two studies establish that the CRTverbal appears to measure a unitary underlying construct, and is correlated with Bayesian reasoning performance. All other measures were also correlated with Bayesian reasoning, and generally with each other (including numerical literacy correlating with the non-mathematical CRTverbal). Stepwise regressions found that Bayesian reasoning was best predicted by visuospatial ability and sets reasoning ability, excluding numeracy, working memory (Experiment 1), and verbal reasoning or pattern extrapolation abilities (Experiment 2). Although CRT remains as a predictor of some utility, these results suggest a different construct may be generating these patterns of correlations.