Failure to Report Effect Sizes

Given the American Psychological Association's strong recommendation to always report effect sizes in research, scholars have a responsibility to provide complete information regarding their findings. The purposes of this study were to (a) determine the frequencies with which different effect s...

Full description

Saved in:
Bibliographic Details
Published in:Health education & behavior 2016-10, Vol.43 (5), p.518
Main Authors: Barry, Adam E, Szucs, Leigh E, Reyes, Jovanni V, Ji, Qian, Wilson, Kelly L, Thompson, Bruce
Format: Article
Language:English
Subjects:
Association (Psychology)
Data
Data processing
Education
Effect Size
Electronic publishing
Health
Health behavior
Health education
Health promotion
Health risk assessment
Medical research
Online Surveys
Periodicals
Promotion
Reporting
Reviews
Social Science Research
Social sciences
Statistical Significance
Citations: Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Given the American Psychological Association's strong recommendation to always report effect sizes in research, scholars have a responsibility to provide complete information regarding their findings. The purposes of this study were to (a) determine the frequencies with which different effect sizes were reported in published, peer-reviewed articles in health education, promotion, and behavior journals and (b) discuss implications for reporting effect size in social science research. Across a 4-year time period (2010-2013), 1,950 peer-reviewed published articles were examined from the following six health education and behavior journals: American Journal of Health Behavior , American Journal of Health Promotion , Health Education & Behavior , Health Education Research , Journal of American College Health , and Journal of School Health . Quantitative features from eligible manuscripts were documented using Qualtrics online survey software. Of the 1,245 articles in the final sample that reported quantitative data analyses, approximately 47.9% (n = 597) of the articles reported an effect size. While 16 unique types of effect size were reported across all included journals, many of the effect sizes were reported with little frequency across most journals. Overall, odds ratio/adjusted odds ratio (n = 340, 50.1%), Pearson r /r 2 (n = 162, 23.8%), and eta squared/partial eta squared (n = 46, 7.2%) accounted for the most frequently used effect size. Quality research practice requires both testing statistical significance and reporting effect size. However, our study shows that a substantial portion of published literature in health education and behavior lacks consistent reporting of effect size.
ISSN:1090-1981
1552-6127
DOI:10.1177/1090198116669521