Normative Data for Semantic Verbal Fluency in the Waranka Population: a Poisson Modeling Approach

Abstract Objective To generate normative data for animal and fruit verbal fluency tasks—total score (TS), cluster size (CS), and number of switches (NS)—in the Waranka minority population of Ecuador. Method A sample of 412 healthy children aged 6–17 from Ecuador was assessed. Semantic verbal fluency scores were normed using multiple generalized linear models (GLMs) with a Poisson distribution. The full model included all main effects and two-way interactions among log-transformed age, log-transformed mean parental education (MPE), sex, and LSBQ score. Variable selection followed an exhaustive search strategy over all 2^p predictor subsets, with model selection based on the Akaike Information Criterion (AIC). Norms were computed through a three-step process: (1) the expected score (${\upmu}_{\rm i}$) was calculated from model estimates (${\upbeta}\square\!\!\!?$); (2) the Poisson cumulative distribution function was used to estimate outcome probabilities; and (3) this value was multiplied by 100 to yield percentiles. Results Age was positively associated with fluency performance across all models. In TS-animals and NS-animals, interactions indicated stronger age-related gains in boys, while higher LSBQ scores were negatively related to performance in boys. For TS-fruits, age and MPE showed positive associations, while LSBQ and female sex showed small negative effects. In NS-fruits, female sex was negatively associated with scores, but this effect reversed with increasing age. TS-fruits showed the best model fit (D2 = 0.29), followed by TS-animals (0.24) and NS-animals (0.16). CS models showed limited predictive value (D2 = 0.10–0.14). Conclusion Age, sex, and bilingual experience jointly shape fluency performance in children, with context-dependent interactive effects.