ANAGRAM EFFECTS IN VISUAL WORD RECOGNITION

P. Courrieu & M. Lequeux / Anagram Effects 12 / 40Response times (and accuracy data, for control) were input into an analysis ofvariance without any correction or exclusion, thus the full random variance ofdata was preserved (this is true for all experiments in this paper). Theindependent variables were the word's frequency of use (frequent vs.infrequent), and the number of lexical anagrams the stimulus had (anagram vs.non-anagram). The F-ratios for the participant population (F1) and the itempopulation (F2) were computed. In addition, we computed Quasi-F ratios (F'formula), which provide a severe test of effects with respect to the participantand item populations simultaneously. Quasi-F ratio tests are suitable tohal-00429184, version 1 - 1 Nov 2009experimental designs using sampled materials that cannot be counterbalanced(Wickens & Kepple, 1983; Winer, 1971, pp. 357-378). In addition, the use ofsuch a severe test, together with a preserved full random variance, preventsany abusive positive conclusion, particularly in the case of a relatively lowexperimental power (only 7 items per condition, due to selection constraints).ResultsThe mean response times and percentages of correct responses are presentedin Table 2. The response times were significantly shorter for the frequent wordsthan for the infrequent ones (F1(1, 23)=18.25, p