An analysis of the effect of presentation mode is presented in
Supplementary Materials S1. Briefly, there were no significant effects of presentation mode or interactions between presentation mode and set size. For the current analysis, we report only the data from the sequential presentation mode. We included all datasets in this condition. Examination of the eye tracking data revealed that, on average, approximately 20 saccades were made on each trial (
M = 19.87,
SD = 8.33, minimum = 1, maximum = 51).
Owing to the small sample size, we ran linear mixed effects model in R version 4.2.1 (
R Core Team, 2019) using the lmerTest package (
Kuznetsova, Brockhoff, & Christensen, 2017), which applies Satterthwaite's method to estimate degrees of freedom and
p values for the overall effect of serial position. The model was run on each set size to examine the effect of serial position after controlling for the random effect of participant. Serial position was included as a fixed effect and we included participant ID as a random effect.
1 Bonferroni–Holm corrected post hoc contrasts of the estimated marginal means were carried out to examine any significant effects, using the emmeans package (
Lenth, 2022).
For imprecision (
Figure 2A), a significant effect of serial position was observed at set size four,
F(3, 9) = 4.96,
p = 0.027. The effect of serial position was not significant at set size two,
F(1, 3) = 0.37,
p = 0.587, or set size three,
F(2, 6) = 4.33,
p = 0.069. Post hoc pairwise comparisons between serial positions revealed a significant difference between the first and second item at set size four (
p = 0.037). No other comparisons were significant (
p ≥ 1.000). For the probability of reporting the target (
Figure 2B), no significant effects of serial position were observed at set sizes two,
F(1, 3) = 0.55,
p = 0.513, three,
F(2, 6) = 1.12,
p = 0.387, and four,
F(3, 9) = 1.22,
p = 0.357. For the probability of misbinding (
Figure 2C), a significant effect of serial position was observed at set size four,
F(3, 9) = 4.67,
p = 0.031. The effect of serial position was not significant at set sizes two,
F(1, 3) = 2.73,
p = 0.197, and three,
F(2, 9) = 0.06,
p = 0.946. Post hoc pairwise comparisons between serial positions revealed no significant differences between items at set size four (
p ≥ 0.145). For the probability of guessing (
Figure 2D), the effect of serial position was not significant at any set size; set size two,
F(1, 3) = 0.96,
p = 0.400; set size three,
F(2, 9) = 0.35,
p = 0.715; set size four,
F(3, 12) = 0.62,
p = 0.617.