I Agreed with What? Memory for Simple Argument Claims

Abstract

Three experiments were conducted to examine how precisely readers recall the claims of arguments that they have just read. Participants read simple, 2-clause arguments such as, “The U.S. is right to intervene in other countries' affairs because local events can catastrophically impact the entire world.” Participants then evaluated each argument with respect to agreement or fit and recalled either the claim or the complete argument. The first 2 experiments found that readers have difficulty precisely recalling the main predicate of the claim (e.g., “should intervene” or “is right to intervene”) as accurately as they do the theme of the argument. Furthermore, a comparison with predicates of short narrative statements indicates that this recall difficulty is specific to argument predicates. Experiment 3 found that skilled readers' recall was more accurate than less-skilled readers. It also showed that the predicate recall problem occurred with both skilled and less-skilled readers and that it is related to one's ability to detect poorly formed arguments.

Notes

¹The focus of follow-up analyses are on the verbatim analysus because it is the most stringent criterion and most apt to show subtle differences.

²We also conducted a regression analysis using the verbatim recall data to assess the impact of recall skill on flawed judgment detection. The criterion variable was the judgment score averaged across good and unwarranted judgments for each participant. We entered the Nelson–Denny score in the first step to control for reading skill. The average recall scores for the predicate were then entered in the second step. The regression equation significantly predicted flawed judgment performance, F(2, 78) = 6.17, MSE = .023, p < .01, R ² = .137. Predicate recall scores accounted for a significant change in R ² after controlling for Nelson–Denny, F(1, 78) = 6.01, p < .02, R ² = .066. The B for Nelson–Denny in the full model was .145 (t = 1.45, p = .216), and the B for predicate recall in the final model was .284 (t = 2.45, p < .02). Thus, predicate recall ability significantly predicted flawed judgment performance over and above the effect of reading skill.