McRae and Cree (2002), and Krauth-Gruber et al. (2004). Figure 7.1 presents a representative protocol coded with this scheme (also see Figure 7.2 later).

In analyzing the results, few differences of theoretical importance occurred at the level of the 45 specific codes. Instead, the interesting results appeared at the level of the general coding categories, aggregating across the specific codes within them. Thus, we only report results from the general level. To establish reliability, a second judge coded the statements for a single participant's nine concepts, and agreed with 95% of the codes given by the judge who coded all of the protocols. Similar levels of agreement using this scheme were reported in Wu and Barsalou (2004), where reliability averaged 90%.

Table 7.1 presents the proportions of protocol statements in the four general coding categories of interest: taxonomic, entity, setting/event, and introspective properties. For each of the 21 participants, the average proportion of properties was computed for each of the 4 general coding categories, once each across the 3 concepts for each of the 3 concept types. These proportions were subjected to an arcsin transformation that normalized variance (Winer, 1971), and then submitted to a mixed ANOVA having context, concept abstractness, and general coding category as factors. Unless noted otherwise, all reported tests were significant at the .05 level. Because the tests assessed a priori predictions, post hoc corrections were not employed. The large size of most F values further reduces the probability of a Type I error. MSEs are reported in arcsin units.

Overall Differences in Content. The proportions for the four general coding categories differed ^(3,57) = 160.13, MSE = .02 arcsin). As Table 7.1 illustrates, setting/event properties were produced most often in the protocols. Half of the properties produced, .50 overall, described aspects of settings and events. Notably, introspective properties were next most frequent (.24), followed by entity properties (.21), and taxonomic categories (.05). In statistical tests of the adjacent differences, setting/event properties were

Position Latency Analysts 1


Analysis. 2

Position Latency Analysts 1

0 0

Post a comment