ERIC Number: ED448204
Record Type: RIE
Publication Date: 2000-Nov
Reliability Generalization: The Importance of Considering Sample Specificity, Confident Intervals, and Subgroup Differences.
Onwuegbuzie, Anthony J.; Daniel, Larry G.
The purposes of this paper are to identify common errors made by researchers when dealing with reliability coefficients and to outline best practices for reporting and interpreting reliability coefficients. Common errors that researchers make are: (1) stating that the instruments are reliable; (2) incorrectly interpreting correlation coefficients; (3) not reporting reliability coefficients for their own sample; (4) conducting tests of statistical significance on reliability coefficients; and (5) failing to report the reliability of difference scores when examining gain scores. It is recommended that researchers report reliability coefficients for their own data and that they interpret confidence intervals around reliability coefficients, considering that reliability coefficients should be reported not only for the full sample at hand but also for each subgroup. A heuristic example is used for the two-sample case (i.e., t-test) to illustrate how comparing subgroups with different reliability coefficients can affect statistical power. (Contains 3 tables and 36 references.) (Author/SLD)
Publication Type: Reports - Descriptive; Speeches/Meeting Papers
Education Level: N/A
Authoring Institution: N/A
Note: Paper presented at the Annual Meeting of the Mid-South Educational Research Association (28th, Bowling Green, KY, November 17-19, 2000).