NotesFAQContact Us
Collection
Advanced
Search Tips
ERIC Number: ED449221
Record Type: Non-Journal
Publication Date: 2001-Feb-1
Pages: 14
Abstractor: N/A
Reference Count: N/A
ISBN: N/A
ISSN: N/A
Some "Modern" Statistics: A Primer and Demonstration.
Locker, Lesley Ann
Researchers in behavioral science have traditionally used "classical" statistics (e.g., mean and standard deviation) in analyzing data and reporting the results of their studies. However, it has been argued that classical statistical methods do not always represent the population well when analyzing sampling data, resulting in reduced statistical significance for many studies. Problems tend to arise when outliers (unusual scores) are drawn from a sample of the population, and distributions are skewed or heavy-tailed. The most common "modern" methods of statistical analysis are "Winsorized" (named after the statistician Charles Winsor) and "Trimmed" means. Both of these modern methods censor the outlying scores of the sample to allow for the mean to characterize the population more accurately. Most researchers, however, are still unaware or have limited knowledge of modern statistics and their benefits. Perhaps new awareness can be attained through a more concrete definition of the differences between "classical" and "modern" statistics. Sole reliance on "classical" methods will continue to reduce the number of statistically significant findings by researchers. (Contains 1 table and 11 references.) (Author/SLD)
Publication Type: Reports - Descriptive; Speeches/Meeting Papers
Education Level: N/A
Audience: N/A
Language: English
Sponsor: N/A
Authoring Institution: N/A