2 b, the difference scores (concrete minus abstract) were significant in the predicted direction for eight of the nine measures ( p < 0.05, two-tailed; cognitive factors reached marginal significance in the predicted direction, p < 0.10).
The average correlation between the dualism scale and the framing effect was significantly negative in Experiment 1, r (41) = −0.38, p = 0.013, and failed to reach significance in Experiment 3, r (37) = 0.34, p = 0.16.
Follow-up analyses conducted separately on the two sets of psychological measures showed that this marginally significant effect on psychological attributions occurred because concrete items were rated significantly higher than abstract items on the Psychological Set 2 measures ( M = 6.74, SD = 0.91 vs.