BERD team helps Illinois investigators better understand effect size and power analysis
12/2/2021 9:31:08 AM
Webinar offers solutions to common issues in data analysis
The IHSI Biostatistics, Epidemiology, and Research Design (BERD) team hosted its second webinar, “Understanding Effect Size and Power Analysis,” on Nov. 17. The recording is now available to view.
The first BERD webinar, “Towards understanding the p-value: Has it lost its significance?,” introduced problems with the use of p-values, and offered solutions for overcoming these issues in data analysis. One recommended solution is to use the p-value to supplement effect sizes, which are central to quantitative health studies. Effect sizes are almost always presented in study results under different guises, such as differences of means, risk ratios, or correlations. Additionally, effect size is an ingredient in power analysis, an essential study planning ingredient that affects study objectives, feasibility, and budget.
Considering different forms of effect sizes
In this webinar, Aman Kaur, BERD Research Biostatistician, discussed the different forms of effect sizes, considerations for their selection, interpretation, and recommendations for their use. "Effect sizes tell us the magnitude of the difference between groups. These must be mentioned in the research papers to show the true difference in the population,” said Kaur.
Kaur explained that there are several effect sizes available based on study designs. Though Cohen’s d has been the most popular one with conventional standards for interpretation (0.2= small effect; 0.5= moderate effect; 0.8=large effect), she suggests there is little empirical justification for these standards.
“One should always take caution in interpreting effect sizes within the given context of the study and not use conventional cut-offs," said Kaur.
Power analysis as a planning tool
Jesus Sarol, BERD Senior Research Biostatistician, described how power analysis is conducted by identifying the main elements and determining the appropriate module for deriving any of these elements. "Power analysis is used to assess study feasibility, appropriateness of statistical analysis, meaningful effect sizes and likelihood of attaining research objectives,” said Sarol. “It is effectively a planning tool. Performing a power analysis is a way of ensuring that one has thought through every aspect of the study and the statistical analysis before conducting the study."
The webinar concluded with a demonstration of power analysis software. Attendees learned that power analysis is a planned activity. Careful execution of power analysis will strengthen the justification for your study proposal, eventually leading to a better study.
Learn more about IHSI’s BERD team and how they help researchers design studies and enhance data collection, management, and analysis for health-related research projects. Subscribe to the IHSI e-newsletter to learn about upcoming webinars hosted by the BERD team and teams from other IHSI cores aimed to help health sciences researchers increase impact at each stage of the research cycle.