The conventional versus a constructionist Scratch programming and first-year students' achievements in higher education classes: experimental data.

Globally, learning or teaching the first programming (popularly called CS1) remains a significant educational challenge. Indicators such as CS1 students' engagement, failure and attrition rates, and lack of diversity, continue to show the need for innovating the learning or teaching of novice computer science students. To ease initiating novices to programming, Scratch, a visual programming language, has become a staple of K-12 CS1 classes. 

As outcomes of a research project aiming to explore a constructionist Scratch pedagogy with novice CS students in higher education, we present these datasets. In the research lasting two successive academic sessions, we conducted two quasi-experimental studies involving four intact CS1 classes in selected public polytechnic in the north central Nigeria. In each study, we randomly assigned the classes to the experimental and control groups, constituting the constructionist Scratch and the conventional CS1 classes, respectively.

Instruments for collecting data include a student profile questionnaire, a pretest, and posttest. Sequel to ethical clearance and permission from the selected schools, we conducted each study during the first semester of each academic session, in the first seven to eight weeks. During the first to second week, we administered students who consented to take part with the questionnaire and the pretest. Learning or teaching in the two classes lasted six weeks. Then both classes took the posttest. An independent CS educator who is not part of this research marked all the achievement tests, following a rubric prepared by the first author. To strengthen the research design and the possibility of arriving at valid causal evidence, we employed a Coarsened Exact Matching (CEM) algorithm to generate matched samples of experimental and control data, which we used in the analysis.

Data presented here includes the raw, unmatched and matched experimental datasets from both studies. A researcher can make use of the data:

To explore if some background variables not addressed in the original research may moderate CS1 students' achievements. For instance, their prior achievements in mathematics, physics, or English.

To uncover some interesting patterns using machine learning algorithms.

To validate the outcome of the original experiment by using the unmatched, matched or newly generated matched samples.

The authors welcome further research collaborations in using the data or the accompanying research instruments.

Enable GingerCannot connect to Ginger Check your internet connection
or reload the browserDisable in this text fieldRephraseRephrase current sentence4Edit in Ginger×