Statistical treatment of looking-time data.

Abstract

Looking times (LTs) are frequently measured in empirical research on infant cognition. We analyzed the statistical distribution of LTs across participants to develop recommendations for their treatment in infancy research. Our analyses focused on a common within-subject experimental design, in which longer looking to novel or unexpected stimuli is predicted. We analyzed data from 2 sources: an in-house set of LTs that included data from individual participants (47 experiments, 1,584 observations), and a representative set of published articles reporting group-level LT statistics (149 experiments from 33 articles). We established that LTs are log-normally distributed across participants, and therefore, should always be log-transformed before parametric statistical analyses. We estimated the typical size of significant effects in LT studies, which allowed us to make recommendations about setting sample sizes. We show how our estimate of the distribution of effect sizes of LT studies can be used to design experiments to be analyzed by Bayesian statistics, where the experimenter is required to determine in advance the predicted effect size rather than the sample size. We demonstrate the robustness of this method in both sets of LT experiments.

Infants’ sensitivity to rhyme in songs

“…The t-test, in contrast, aggregates over songs and trials. Accounting for nuisance variance in the data may be generally important in infant looking time experiments, as the effect sizes in such experiments tend to be small (Bergmann et al, 2018; but see again: Csibra et al, 2016). In the present study, the observed effect size (d z = 0.28) was substantially smaller than the anticipated effect size (d z = 0.5).…”

“…The linear mixed effect model was fit onto Box-Cox transformed looking times (lambda = 0.32), because the residuals of the same model with untransformed data were not normally distributed, as assessed through visual inspection (see also Csibra, Hernik, Mascaro, Tatone, & Lengyel, 2016 for recommendations on why to log-transform infant looking time data). The fixed factors of the model were 1) condition (rhyme, non-rhyme), coded as orthogonal contrasts; 2) test trial number (1-14), coded as a linear polynomial; and 3) the interaction of condition and test trial number.…”

Infants' visual sustained attention is higher during joint play than solo play: is this due to increased endogenous attention control or exogenous stimulus capture?

“…Results show that all look durations are heavily positively skewed (see Figures 4c-e), as is universally observed in look duration data(Csibra, Hernik, Mascaro, Tatone, & Lengyel, 2016), and so a log transform was applied. Results show that all look durations are heavily positively skewed (see Figures 4c-e), as is universally observed in look duration data(Csibra, Hernik, Mascaro, Tatone, & Lengyel, 2016), and so a log transform was applied.…”

Predictive action tracking without motor experience in 8-month-old infants

“…2). Because looking times tend to be distributed log-normally across participants (Csibra, Hernik, Mascaro, Tatone, & Lengyel, 2016), the data were log-transformed prior to parametric analyses. A repeated measures ANOVA with continuation (correct vs. incorrect) and orientation (upright vs. inverted) as within subjects factors demonstrated a main effect of continuation, F (1, 23) = 5.702, p  = 0.026, ${\eta }_p^2$ = 0.199.…”