NotesFAQContact Us
Search Tips
ERIC Number: ED448188
Record Type: Non-Journal
Publication Date: 2000-Nov
Pages: 16
Abstractor: N/A
Reference Count: N/A
Are All Effect Sizes Created Equal?
McLean, James E.; O'Neal, Marcia R.; Barnette, J. Jackson
This study compared effect sizes applied to raw, scaled, and normal curve equivalent (NCE) data. Recommendations for the interpretation of effect sizes vary. For example, some authors suggest that an effect size below 0.50 is small, between 0.50 and 1.00 is moderate, and above 1.00 is large. These are products of the criterion formally used by the U.S. Department of Education's Joint Dissemination Review Panel and the Program Effectiveness Panel. it is clear from the context of these articles that it is assumed that they were dealing with raw scores or scaled scores, not NCEs. NCE scores for individual students and, particularly, mean NCE scores for schools would not be expected to change from year to year without some type of intervention. This study computed effect sizes for the raw, scaled scores, and NCEs by school for grades 4, 6, and 8 on a national norm-referenced test for 749, 574, and 464 schools respectively representing 120,149 students. The results show that, as expected, the effect sizes for NCE scores were lower than those for raw and scaled scores. These results suggest that when rules-of-thumb for effect sizes are presented, they should take into account the type of metric on which it is being applied. (Contains 3 figures, 4 tables, and 23 references.) (SLD)
Publication Type: Reports - Research; Speeches/Meeting Papers
Education Level: N/A
Audience: N/A
Language: English
Sponsor: N/A
Authoring Institution: N/A