Water Entry Matters: Toddlers’ Avoidance of Bodies of Water

Eighty-five toddlers participated. Parent interviews provided structured developmental data. Two had never crawled, and for two, crawling onset (defined as the date of ten consecutive belly or hands and knees cycles) was unreported. All parents supplied walking onset, marked by ten independent steps. Walking experience was calculated as the period from walking onset to test date; total locomotor experience was counted from the earliest self locomotion to testing. For aquatic exposure, 27 were tested at swim lesson pools, ensuring familiarity. Twenty had no swim lessons, 15 had attended less than ten, and 48 had attended ten or more sessions. The Ethics Committee for Research of the Faculty of Human Kinetics, University of Lisbon, approved procedures (CEIFMH 27/2023).
Procedure: Each child underwent one randomized trial each on a water cliff and slope. A safety harness fixed to a rope kept the child secure, limiting immersion to the chin. After acclimatization, parents were placed opposite and signaled the child. The experimenter threw a ball (1m from pool edge in cliff, 2m from entry in slope). The toddler was positioned standing at the edge, facing their parent. They were asked to retrieve and deliver the ball. Trials ended after 150s without entry, if the child entered water (fell, stepped, jumped—cliff: suspended by rope; slope: reached submersion point), if the toddler left and couldn’t return, or began crying and couldn’t be calmed (excluded).
Data coding: All trials were videotaped. Analysts coded behaviors for Avoidance (avoider/non-avoider) in cliff and slope conditions. Non avoiders were suspended at the cliff or reached submersion in the slope; avoiders refrained from both for 150s or moved away and could not be coaxed back.
Analysis: Data normality was checked with Shapiro–Wilk, showing normal distributions. Pearson’s Chi-square tested effects of sex, trial order, and pool familiarity on avoidance. Binomial logistic regressions evaluated impacts of age, crawling, walking, total locomotor experience (TL.exp), and baby swimming sessions (BSS) on avoidance at the cliff and slope. When multiple predictors were significant, model fit and out-of-sample error were compared using Akaike’s Information Criterion (AIC); models ΔAIC≤2 had strong support. McNemar’s test compared avoidance rates across conditions; Fisher’s Exact test evaluated avoidance by swim session attendance (<10 vs. ≥10).