On this page, we present the material for several experiments we conducted to evaluate program comprehenion in the context of feature-oriented software development. Additionally, we present "meta-experiments" to help the empirical-software-engineering community to improve the quality of controlled experiments measuring program comprehension.
Currently, we evaluate whether we can use functional magnetic resonance imaging to measure program comprehension.
We asked program-committee and editorial-board members of major software-engineering venues on their opinion regarding empirical studies in software engineering, especially their view on validity and replication.
In a family of three experiments, we evaluated whether background colors improve program comprehension in preprocessor-based software product lines.
In a controlled experiment, we evaluated whether software measures can be used as indicators for program comprehension.
Programming Experience is a major confounding parameter for program-comprehension experiments. To measure it, we are developing a questionnaire and can show first results.
To reliably measure program comprehension with controlled experiments, we need to control confounding parameters. To support researchers, we developed a list and describe control techniques.
To support designing, conducting, and replicating experiments, we developed the tool PROPHET.
To evaluate whether physical separation of concerns improves program comprehension compared to virtual separation of concerns, we are currently conducting experiments. We the experimental setting and present first results.
A visualization concept to help keeping an overview of a feature-oriented software system.
We summarized the most important material of all experiments in one zip-file: Summary
Responsible for content and maintenance: Janet Siegmund (siegj@hrz.tu-chemnitz.de)