Lertap 5's main measure of test and scale reliability is coefficient alpha, also known as Cronbach's alpha. There are many other measures of reliability; the preceding topic discusses some of them. When a test with a cut-score is used, measures of decision consistency will often become primary indicators of test quality -- these measures are discussed in another topic.

 

The text by Meyer (2010) is a highly-recommended reference in all of these areas.

 

Coefficient omega, also known as McDonald's omega, is yet another measure of reliability, not discussed in Meyer's text. It has been shown that omega is a superior measure of reliability as it makes fewer assumptions; this is now widely acknowledged, but practical studies, such as this one, have found that alpha and omega will be generally not be noticeably different.

 

The "Item scores and correlations" option in Lertap will produce the "IStats" report. An estimate of omega computed using the "closed-form" method will be found towards the end of the report. An example is found in Appendix D of this working paper.

 

Those interested in omega might also want to make use of a special Lertap macro, "Omega1". This macro will create a special 'csv' data file and a small text file with several lines of R code designed for use with the Psych toolbox from CRAN.