Developmental trajectories of accuracy and type of errors in Fluid intelligence assessment as detected by a new digital tool: MatriKS.

Fluid intelligence (FI) is the ability to think logically and solve problems in new situations. It is generally assessed through nonverbal reasoning tests (e.g.,Raven’s Matrices). Studies assessing FI through standard tests highlighted that this skill improves during development. Moreover, prior research showed that the matrices’ complexity is influenced by several factors - such as the number of elements, perceptual characteristics, rules number, and complexity. Yet, to date few evidence exists on changes in different types of errors during development. This study aims to evaluate the developmental trajectories of MatriKS, a new digital tool for the evaluation of FI. We analysed the total accuracy and the different types of errors (i.e., repetition, wrong principle, difference, incomplete correlate). MatriKS was administered to N=590 typically developing participants (males = 311, females = 279) aged 4-11 years old (M = 8.39, SD = 2.19). Results of generalised linear model (glm) on accuracy showed that children’s performance significantly improves with age (z = -29.06, p<0.001). Results of glm on error types showed an overall decrease of errors number with age (z = -23.58, p<.001). To better investigate the relation between changes in the different types of errors and age, we divided the sample in 2 years-step age groups and performed an ANOVA. Results showed significant main effects of error type (F(5,561)=30,91, p<0.001) and age (F(4,561)=89.64, p<0.001), as well as a significant interaction (F(11,1300)=5.94, p<0.001). Specifically, the number of errors significantly decreases with age in all the distractor types, but not in “Difference”. In conclusion, MatriKS proves to be a tool capable of detecting differences based on age, both on accuracy and error types. Information about the characteristics of errors and how they change with age is pivotal for planning targeted and individualized interventions in several clinical populations.