We analyzed the data from *n* = 85 participants who completed the study using jamovi. The jamovi file with labeled variables and results is provided on this OSF page, along with a .csv copy of the data.
We first tested whether knowledge ratings differed between conditions (Gettier case, ignorance control, or knowledge control) for the Darrel vignette using a One-Way ANOVA (Welch's). Knowledge ratings did not significantly differ between conditions for the Darrel vignette, *F*(2, 54.1) = 1.25, *p* = .294. Because the omnibus test was not significant, we did not conduct any follow-up post-hoc tests.
Next, we tested whether knowledge ratings differed between conditions for the Gerald vignette by first conducting a One-Way ANOVA (Welch's). Knowledge ratings significantly differed between conditions for the Gerald vignette, *F*(2, 53.7) = 9.18, *p* < .001. Levene's Homogeneity of Variances Test was significant (*F* (2, 82) = 6.63, *p* = .002) so we conducted Games-Howell post-hoc tests rather than using Tukey HSD. The knowledge ratings for the Ignorance condition (*M* = 19.8, *SD* = 33.1) were significantly lower than the ratings for either the Gettier case condition (*M* = 45.5, *SD* = 43.4; *t*(48.5) = 2.43, *p* = 0.048) or the Knowledge control condition (*M* = 60.4, *SD* = 39.9; *t*(56) = 4.25, *p* < 0.001).
Finally, we tested whether knowledge ratings differed between conditions for the Emma vignette by conducting a One-Way ANOVA (Welch's). Knowledge ratings did not significantly differ between conditions for the Darrel vignette, *F*(2, 51.1) = 0.58, *p* = .562. Because the omnibus test was not significant, we did not conduct any follow-up post-hoc tests.
----------
The jamovi project (2020). jamovi. (Version 1.6) [Computer Software]. Retrieved from https://www.jamovi.org.