userfriendlyscience.com

Scientific research made accessible

      Codecov
Home | Documentation | Download | Reliability | Diamond Plots

The userfriendlyscience R package offers a number of functions to make computing a number of reliability estimates very easy. Please see https://userfriendlyscience.com for installation instructions.

These functions are explained in a number of papers. The first one is open access and about Coefficient Alpha ("Cronbach's Alpha"), Omega, and the Greatest Lower Bound. The second one is also open access and is about test-retest reliability. Both of these are in a special issue on methodology in health psychology of the European Health Psychologist. A third paper embeds these points in a broader framework: this open access paper in Health Psychology Review also covers factor analysis and provides guidelines for authors, reviewers and editors regarding scale quality and what should be reported. If you use an alternative to coefficient Alpha from the userfriendlyscience package, please cite either the first or this last paper, and if you use a test-retest coefficient function from userfriendlyscience, please cite the second one.

Finally, an overview paper was published in Psychological Methods by Daniel Nash. He hosts a self-archived version of the manuscript at ResearchGate. This manuscript is an excellent, comprensive yet concise and accessible introduction to the problems with alpha, so it's a great read after the more accessible introductions linked to above. It also introduces coefficient H, another alternative to alpha apart from omega and the GLB.

This page describes a number of the functions to compute coefficient alpha, omega, the GLB, and coefficient H, as well as test-retest reliability coefficients.


Some examples of these functions:
### This will prompt the user to select an SPSS file (see the article for ### an explanation, or just use ?scaleStructure in R to see the help). scaleStructure(); ### Load data from simulated dataset testRetestSimData (which ### satisfies essential tau-equivalence). data(testRetestSimData); ### Select some items in the first measurement exampleData <- testRetestSimData[2:6]; ### Use all items to compute the glb and omega scaleStructure(dat=exampleData, ci=FALSE); ### Use a selection of two variables scaleStructure(dat=exampleData, items=c('t0_item2', 't0_item4'), ci=FALSE); ### Note that 'ci = FALSE' can be omitted to compute confidence intervals ### for scales of three items or more. ### It also contains functions to compute the test-retest reliability. ### The first column is the true score, so it's excluded in this example. exampleData <- testRetestSimData[, 2:ncol(testRetestSimData)]; ### Compute test-retest alpha coefficient testRetestReliability(exampleData);

If you have any questions, you can contact me through https://behaviorchange.eu.