A repeated measures ANOVA also revealed a marginally significant three-way interaction effect between frequency, length and regularity, with F(1,20) = 4.57; p = 0.045, η 2 p = 0.186, in which, although the obtained p-value is close to the significance level of 0.05, we found a large effect size that could explain the interaction of these three psycholinguistic factors on Gaze Duration.
The fit of a random intercept model across volunteers indicated statistical significance, according to the Wald chi-squared test, for frequency (low-frequency - high-frequency, linear estimate ± standard error 0.2896±0.1371, Χ 2 = 4.4634, p = 0.035, odds 1.336), for length (long-short, 0.6256±0.1375, Χ 2 = 20.7091, p < 0.0001, odds 1.869), for the interaction between regularity (irregular-regular) and frequency (-0.7527±0.2743, Χ 2 = 7.5305, p = 0.006, odds 0.4711) and marginal significance for the three-way interaction between length, frequency and regularity (-1.0561±0.5482, Χ 2 = 3.7114, p = 0.054, odds 0.3478).
The effect size was large, with odds of 2.077 and 2.861, respectively; the interaction, however, did not reach statistical significance (0.2104±0.2340, Χ 2 = 0.8081, p = 0.37).