The effect of developers’ General Intelligence on the Understandability of Domain Models: an Empirical Study

Abstract

The discipline of software engineering has long studied the understandability of domain models, but always focusing on the semantic and notational characteristics of these models. However, as some authors point out, understandability is a cognitive process, where many human factors of the developers themselves are involved. In this sense, one of the human factors most studied by cognitive psychology is the Intelligence, considered as a set of skills that give each person abilities among which stand out comprehensibility and problem solving, both associated with the understandability process. In this paper we focus on Spearman’s Bifactorial Theory that proposes the use of D48 test, to obtain the so-called General Intelligence that represents a factor underlying specific mental abilities. This work proposes a theoretical model that has guided an empirical study with 102 subjects from the University of Alicante, we measure their general intelligence using the D48 test, and then these subjects perform a set of UML domain model understandability tasks, both in comprehensibility and problem solving. At this point, we also study the impact that the model understandability performance has on the final intention to adopt the model, based on the UMAM-Q test. Once the data were obtained, we applied a two-way analysis of variance, and the study obtained statistically significant results confirming that: 1) subjects with higher intelligence perform obtain better results on model understandability performance, and 2) subjects with higher model understandability also have higher compatibility and intention to adopt the model.