NotesFAQContact Us
Collection
Advanced
Search Tips
ERIC Number: ED315439
Record Type: Non-Journal
Publication Date: 1988-Nov
Pages: 19
Abstractor: N/A
Reference Count: N/A
ISBN: N/A
ISSN: N/A
Alternative Logics for Estimating whether Research Results Will Generalize.
Campo, Stephanie F.
Three procedures for evaluating the sampling specificity of results are reviewed. These procedures are Tukey's jacknife technique, Efron's bootstrap technique, and cross-validation methods. The jacknife technique uses different subsamples derived from the original total data set to provide empirical estimates of the generalizability of effect sizes. The bootstrap technique estimates the statistical accuracy of effect size estimates and creates a megafile by copying the original data sample over and over again many times. The researcher then randomly selects a given number of bootstrap samples of size "n" from the megafile. The effect size is computed for each bootstrap sample; these correlation coefficients are treated as a distribution from which statistical estimates of result stability are derived. Cross-validation methods involve the arbitrary splitting of a sample. Prediction equations developed for each group are crossed so that each group will use the other group's prediction equations. A small data set is used to illustrate in more detail how the cross-validation procedure is performed and interpreted. Two data tables and sample Statistical Analysis System commands are provided. (TJH)
Publication Type: Reports - Evaluative; Speeches/Meeting Papers
Education Level: N/A
Audience: N/A
Language: English
Sponsor: N/A
Authoring Institution: N/A