B.1.5.2. Results

The number of excluded values averaged 6 % of trials per participant. First, a general ANOVA was performed on correct RTs, with target size (small, medium and large) and relevance (random, small-relevant, large-relevant) as within-subject factors. The effect of relevance was not significant (F(2,34) = 1.47, MSE = 18,064, p > .24). The target size effect was significant (F(2,34) = 6.14, MSE = 47,730, p < .006), as well as the interaction between both factors (F(4,68) = 41.26, MSE = 464,394, p < .001).

Then, we computed planned comparisons with bilateral Student t-tests, and adjusted the alpha threshold according to the Bonferroni-type procedure proposed by Larzerele and Mulaik (1977). In the random condition, RTs for the small target (913 ms) were significantly longer than those for the medium target (875 ms, t(17) = 2.52; t'_crit = 2.46; p < .001), which did not differ significantly from those for the large target (840 ms, t(17) =1.35; t'_crit = 2.11; p > .096). In the large-relevant condition, RTs for the small target (1081 ms) were significantly longer than those for the medium target (938 ms, t(17) = 5.42; t'_crit = 3.06; p < .001), which were longer than those for the large target (710 ms, t(37) = 6.07; t'_crit = 3.12; p < .001). In the small-relevant condition, RTs for the small target (778 ms) were significantly shorter than those for the medium target (888 ms, t(17) = 3.90; t'_crit = 2.65; p < .001), which were shorter than those for the large target (1045 ms, t(17) = 4.52; t'_crit = 2.90; p < .001). The effects of relevance (i.e. mean RTs in random condition minus mean RTs in relevant condition, see Exp. 2) were also significant for each relevant target size: large targets (129 ms, SE: 24; t(17) = 5.42; t'_crit = 3.06; p < .001) and small targets (133 ms, SE: 33; t(17) = 4.08; t'_crit = 2.79; p < .001).

Finally, vincentizations and paired-comparisons for short and long RTs quintiles were also computed, as in Experiments 1 and 2. Results were presented in Table 1. For the random condition, the only comparison that remained significant was between small and medium targets at short quintiles. In the large-relevant condition, the comparisons were significant between small and medium at short and long quintiles, but only at long quintiles between medium and large. In the small relevant condition, the small-medium comparison was significant only at short quintiles, while the medium-large comparison was significant at both short and long RT quintiles. These results should be considered in the light of the smaller number of trials by condition by quintile (relative to Exp. 1), and the smaller number of participants (relative to Exp. 2).