NotesFAQContact Us
Collection
Advanced
Search Tips
Peer reviewed Peer reviewed
Direct linkDirect link
ERIC Number: EJ1015749
Record Type: Journal
Publication Date: 2013-Sep
Pages: 9
Abstractor: As Provided
Reference Count: 17
ISBN: N/A
ISSN: ISSN-1531-7714
Normality of Residuals Is a Continuous Variable, and Does Seem to Influence the Trustworthiness of Confidence Intervals: A Response to, and Appreciation of, Williams, Grajales, and Kurkiewicz (2013)
Osborne, Jason W.
Practical Assessment, Research & Evaluation, v18 n12 Sep 2013
Osborne and Waters (2002) focused on checking some of the assumptions of multiple linear regression. In a critique of that paper, Williams, Grajales, and Kurkiewicz correctly clarify that regression models estimated using ordinary least squares require the assumption of normally distributed errors, but not the assumption of normally distributed response or predictor variables. They go on to discuss estimate bias and provide a helpful summary of the assumptions of multiple regression when using ordinary least squares. While we were not as precise as we could have been when discussing assumptions of normality, the critical issue of the 2002 paper remains--researchers often do not check on or report on the assumptions of their statistical methods. This response expands on the points made by Williams, advocates a thorough examination of data prior to reporting results, and provides an example of how incremental improvements in meeting the assumption of normality of residuals incrementally improves the accuracy of confidence intervals. (Contains 1 table, 7 figures, and 5 footnotes.)
Dr. Lawrence M. Rudner. e-mail: editor@pareonline.net; Web site: http://pareonline.net
Publication Type: Journal Articles; Opinion Papers; Reports - Evaluative
Education Level: Higher Education; Postsecondary Education
Audience: N/A
Language: English
Sponsor: N/A
Authoring Institution: N/A