R open source statistical software continues to make advances in the open-source tools it offers market researchers.
1. Conducting transparent analyses
Efficiently producing transparent analyses is a perennial challenge. This requires computing systems and environments that can efficiently satisfy reproducibility and accountability standards.
R has now developed a system called adapr (Accountable Data Analysis Process in R) that is built on the principle of accountable units.
An accountable unit is a data file (statistic, table or graphic) that can be associated with a provenance, meaning how it was created, when it was created and who created it, and this is similar to the ’verifiable computational results’ (VCR) concept.
Accountable units use file hashes and do not involve watermarking or public repositories like VCRs. Reproducing collaborative work may be highly complex, requiring repeating computations on multiple systems from multiple authors; however, determining the provenance of each unit is simpler, requiring only a search using file hashes and version control systems.
2. Reducing rating scale items without predictability loss
R offers an innovative method for reducing the number of rating scale items without predictability loss.
The “area under the receiver operator curve” method (AUC ROC) is used for the stepwise method of reducing items of a rating scale. RatingScaleReduction R package contains the presented implementation.
Differential evolution is applied to one of the analysed datasets to illustrate that the presented stepwise method can be used with other classifiers to reduce the number of rating scale items (variables).
This approach can be applied to decision making, data mining, machine learning, and psychometrics.
3. Using Bayesian models for selecting variables in linear models
R offers the option to consider objective Bayesian methods for hypothesis testing and variable selection in linear models.
BayesVarSel is an R package that computes posterior probabilities of hypotheses/models and provides a suite of tools to properly summarize the results.
BayesVarSel contains exact algorithms to perform fast computations in problems of small to moderate size and heuristic sampling methods to solve large problems.
The upshot of these developments is growing sophistication in the range of tasks that R can offer a researcher. To learn the core functionality in statistical analysis for market and social researchers, supported by R, please see details of the following course;