Data illustrate correlations between results from individual participants in cross-over experiment P2007 (Smell and Library) conducted by Madeyski, see: [1] Lech Madeyski, Test-Driven Development: An Empirical Evaluation of Agile Practice. (Heidelberg, London, New York): Springer, 2010. Foreword by Prof. Claes Wohlin. If you use this data set please cite: [1] Lech Madeyski, Test-Driven Development: An Empirical Evaluation of Agile Practice. (Heidelberg, London, New York): Springer, 2010. Foreword by Prof. Claes Wohlin. [2] Barbara Kitchenham, Lech Madeyski, Giuseppe Scanniello and Carmine Gravino, 'The importance of the Correlation between Results from Individual Participants in Crossover Experiments' (to be submitted as of 2020).
KitchenhamEtAl.CorrelationsAmongParticipants.Madeyski10`KitchenhamEtAl.CorrelationsAmongParticipants.Madeyski10`: a data frame with 45 rows and 10 variables:
<fct>| ExperimentID: This experiment is the only cross-over experiment in the family of TDD and Pair-Programming experiments conducted by Madeyski, so all values in this column are set to 'P2007'.
<fct> | Participant ID: An identifier for each participant, unique for a specific experiment.
<fct> | Experimental Sequence Group: A (TLSP-TFSP), B (TFSP-TLSP)
<fct> | Software system to develop: Smell (a tool for identifying bad code smells in Java source code through the use of a set of software metrics) or Library (a library application)
<fct> | Time period of the cross-over experiment: 1 or 2
<fct> | Experimental Treatment: Test-First Solo Programming (TFSP) vs Test-Last Solo Programming (TLSP)
<dbl> | Dependent variable: Percentage of Acceptance Tests Passed
<dbl> | Dependent variable: Number of Acceptance Tests Passed Per Hour
<dbl> | Dependent variable: Mean value of Coupling Between Objects (CBO), see CK set of metrics
<dbl> | Dependent variable: Mean value of Weighted Number of Methods in Class (WMC), see CK set of metrics
<dbl> | Dependent variable: Mean value of Response For a Class (RFC), see CK set of metrics
<fct> | Cross-Over Code. This experiment is a simple two-group cross-over experiment with one cross-over code, so all values in this column are set to 'CO1'. However, four-group experiments require a code to identify the linked sequence groups (although that can be deduced from the system used in the first time period). A crossover code is also essential for non-parametric analysis.
KitchenhamEtAl.CorrelationsAmongParticipants.Madeyski10
Run the code above in your browser using DataLab