Friday 5 October 2012

Sample Size and Power-Analysis for Cluster-Randomized and Multi-Site Studies (power-analysis for cluster-randomized and multi-site studies)

Cluster-randomized (or multi-site) studies are studies with a multi-level structure, such as students within classes within schools, or patients within doctors within hospitals.

Unlike simple-randomized trials, where you need to consider only the number of subjects, in cluster-randomized and multi-site trials you need to consider the number of units at each level of the study design. If you ignore these, you might compute power as 90% when the actual value is 50% or even lower.

Cost-effective design.

Your real challenge, however, is not only to compute the correct power, but to design the most cost-effective study.

In simple randomized trials, the only sample size you can manipulate is the number of subjects. By contrast, in a cluster-randomized or multi-site trial you can manipulate (for example) the number of school districts, schools, classes, and students. Or, the number of hospitals, doctors and patients.

Typically, there are many combinations of these numbers that will yield the same power but (in common scenarios) some may cost twice as much as others.  In this course you will learn how to quickly and confidently find the best options, taking into account both cost and practical considerations.

For more details about this course, Sample Size and Power-Analysis for Cluster-Randomized and Multi-Site Studies, please visit at http://www.statistics.com/CT-Cluster/.

In this course Dr. Borenstein will explain how cluster-randomization and/or the use of multiple sites affect power. Then, he will show how to take account of these issues when planning a study.

The course will cover the following topics
·         Basic issues and goals in a power analysis
·         Cluster-randomized (hierarchical) designs
·         Multi-site (matched or randomized block) designs
·         Two-level designs
·         Three-level designs
·         Four-level designs
·         Working with continuous data
·         Working with binary data
·         Working with covariates
·         Finding the most cost-effective design
·         Using the software to create reports, tables and graphs

How the course works
·         Participants can ask questions and exchange comments with Dr. Borenstein via a private discussion board throughout the period.
·         As with all courses at Statistics.com, participate at your own convenience; there are no set hours when you must be online.  We  estimate you will need about 15 hours per week.
·         This course will include lessons posted online, video lectures, PowerPoint presentations, and video demonstrations of the software.

About the software

Dr. Borenstein is the co-author (with Larry Hedges) of CRT-Power, a program developed with funding from the National Institutes of Health (see below for details).

·         Participants will have free access to this program for the duration of the course.

More info about the online course:
http://www.statistics.com/CT-Cluster/

More info about the software:
http://crt-power.com/

About the instructor:
Michael Borenstein is the co-author of Power And Precision, the best-selling program for power analysis, and of Comprehensive Meta-Analysis, the best-selling program for meta-analysis.  He is also the co-author (with Larry Hedges) of the newly released program CRT-Power, the first program ever developed to compute power for a full array of three-level and four-level study designs.

Dr. Borenstein has taught numerous workshops and online courses on power analysis and meta-analysis, and is widely recognized for his ability to communicate difficult concepts in ways that are accessible to researchers as well as statisticians.

For Indian participants statistics.com accepts registration for its courses at special prices in Indian Rupees through its partner, the Center for eLearning and Training (C-eLT), Pune.

For India Registration and pricing, please visit us at www.india.statistics.com.

Call: 020 66009116

Websites:

No comments: