Computational thinking in university students: The role of fluid intelligence and visuospatial ability

97

Non-programming adults in Hungarian higher education.

A cross-sectional correlational and serial-mediation study. Fluid intelligence (Scrambled Adaptive Matrices), visuospatial ability and crystallised intelligence were measured with computerised adaptive tests with estimated reliabilities of .92-.97, and computational thinking with the Computational Thinking Test. Small n and an adult sample; one author is a director of the test company. Its value is that the sample is PROGRAMMING-NAIVE, which isolates what a computational-thinking test measures before anyone has been taught anything.

Fluid intelligence correlates r = .56 with computational thinking — 31% shared variance — and predicts it after controlling for gender, age and visuospatial ability. Visuospatial ability correlates r = .53 but its effect is fully explained by fluid intelligence. Crystallised intelligence is unrelated (r = .13, not significant). The authors' conclusion is the archive-relevant one: programming-naive computational thinkers draw on reasoning ability that does not rely on previously acquired knowledge, and computational-thinking tests and fluid-intelligence tests sample an overlapping set of underlying visuospatial processes. They also report a large male advantage on the CT test (d = 1.03) that shrinks to r = .176 once ability is controlled.

HIGH by construction — this is a study of the ability structure underlying the outcome instrument, and fluid intelligence is among the most heritable traits in psychology.

Replicates and sharpens Román-González et al. (2017) using item-response-theory-calibrated adaptive ability tests in a programming-naive sample.