Identifying Predictors for Code Highlighting Skills : A Regressional Analysis of Knowledge, Syntax Abilities and Highlighting Skills

Autor: Matthias Kramer, Mike Barkmin, Torsten Brinda
Jazyk: angličtina
Rok vydání: 2019
Předmět:
Zdroj: ITiCSE
Popis: Programming comprises a multitude of involved skills and abilities. To assess these, an even larger multitude of tasks exists, ranging from very complex to basal levels. Recent demands call for the assessment of basal skills, with the aim to figure out possible problems of learners more precisely. In order to read and interpret source code sufficiently, students must be able to recognize given concepts in a source code, e.g. to identify the objects and methods in an object-oriented program. Whether this skill relies solely on conceptual knowledge and the abilities in handling syntax is still unclear. This study gives evidence that isolated skills in these areas are outperformed by the interaction of both areas. We developed a test consisting of conceptual questions, fill-in-the-Java-keyword items and highlighting items. We investigated whether the knowledge about object orientation and the skills in handling Java syntax sufficiently predict the ability to recognize concepts in Java source code via a multiple linear regression. We found that the model with interaction term explains the data better compared to the null model, the isolated predictors or combined predictors without interaction term. This leads to the conclusion that even though programming consists of skills in certainly different areas, it is more important to interconnect these areas with each other than to teach them in an isolated manner.
Databáze: OpenAIRE