Biostatistics Resources
Biostatistics Resources
| General Resources | Plan and design stage | ||
| Power and Sample size | Statistical Analysis | ||
| Getting statistical software at OSU | Missing Data | ||
| Miscellaneous Resources | Glossary |
General Resources
- CTSpedia: A Knowledge Base for Clinical and Translational Research
CTSpedia was created as a national effort to collect wisdom, tools, educational materials, and other items useful for clinical and translational researchers and to provide timely and useful advice to clinical and translational researchers with specific problems. - Biostatistics Resources at GraphPad
- Stattrek Tutorials and Resources
Plan and design stage
There are many important factors that investigators need to consider before a research is carried out. A well-designed experiment will not only yield valid results, but also save time and budget.
The following explains important concepts in the plan and design stage:
- Statistical considerations for clinical trials and scientific experiments
Massachusetts General Hospital’s Biostatitsics Center - Course notes on experimental design
Department of Statistics, Yale University - A non-technical guide of experimental design
Lutherie Information Website - Explanation on levels of measurements
Web center for Social Research Metods - Selecting covariates by Vance W. Berger, PhD
Drug Information Association - Confounding variables
Experiment-Resources.com - Confounding factors and their effects
STATS at George Mason University - A more mathematic way to explain confounding
Engineering Statistics Handbook at National Information Technology Laboratory - Confounding and causality
R McNamee, Occup Environ Med 2003;60:227–234 - Strategies to reduce confounding
StatsDirect Limited
Power and Sample size
Recognition of the importance of power analysis and sample size calculation is one important aspect in experimental design. Without correct calculations that depend on the design, sample size may be too high and may waste time and resources for minimum gain, or too low to reach the desired statistical power.
Below is a list of useful online resources for power and sample size calculation:
- A brief introduction to power analysis
Jeremy Miles at RAND Coportation - Web-based java applets for power and sample size calculation
Russell V. Lenth, PhD at Department of Statistics and Actuarial Science, The University of Iowa - Free interactive program for power and sample size calculations
Department of Biostatistics, Vanderbilt University School of Medicine - Summary of power and sample size Programs
Division of Biostatics, Department of Epidemiology and Biostatistics, University of California, San Francisco
Statistical Analysis
This section serves as a guide on how to choose the appropriate statistical test. The first few links provide a summary of available statistical analyses. When in doubt, always consult with a statistician.
- Which statistical test should I use?
Stat Computing, Academic Technology Services, UCLA - Review of available statistical tests
Intuitive Biostatistics, graphpad.com - Summary statistics and exploratory graphs
Documentation for Analyse-it - Examples of summary statistics
Scott Preston, PhD, SUNY Oswego - Reporting Statistical Results in Your Paper
Greg Anderson, PhD, Bates College - Missing data
Richard William, Department of Sociology, University of Notre Dame - Working with missing values
Alan C. Acock, PhD, Department of Human Development and Family Sciences, Oregon State University - A Statistics tutorial on hypothesis tests
StatTrek.com - Some hypothesis testing examples
Brian J. Lopes, PhD student, Department of Statistics, The University of North Carolina at Chapel Hill - Glossary on hypothesis testing
Statistical education through problem solving, The Department of Statistics, University of Glasgow - Longitudinal research
Community-based intervention research group, Florida Internation lUniversity - Mathematical introduction to longitudinal data analysis
Christopher David Desjardins, Chu-Ting Chung, Jeffrey D. Long, University of Minnesota - Lecture notes on survival analysis
John Fox, PhD, Department of Sociology, McMaster University - A “survivable” introduction to survival analysis
Stephen D. Kachma, PhD, Department of Biometry, University of Nebraska–Lincoln - Introduction to logistic regression
Michael T. Brannick, PhD, Psychology Department, University of South Florida - A mathematical interpretation of logistic regression
Jia Li, PhD, Department of Statistics, The Pennsylvania State University - Logistic regression
Jason T. Newsom, Ph.D, School of Community Health, Portland State University
Getting statistical software at Ohio State
The Ohio State University offers a few statistical software for free or at low cost to faculty, staff and students:
- Free JMP (version 9) Windows XP/Vista/7
- Free JMP (version 9) MacOS X1.05 or later
- Free MINITAB (release 16) Windows XP/Vista/7
- SAS Teaching and Research (version 9.2) Windows XP Pro SP2/Vista Workstation
$33 Faculty and staff on Main and Regional campuses, on OSU- and personally-owned machines. Each copy purchased allows an installation at work and at home. - SPSS Statistics (version 19)/Amos 19 Windows XP/Vista/7
Faculty and staff: $32 per installation; students: no charge. - SPSS Statistics (version 19) MacOS X 10.5 or later
Faculty and staff: $32 per installation; students: no charge. - STATA (version 12) Windows XP/Vista/7
Available to faculty, staff, and students on main and regional campuses and on home computers. Licenses purchased with OSU funds remain with OSU. To purchase, call 800-782-8272 or go to the StataCorp site. - STATA (version 12) MacOS X 10.5 and later
Available to faculty, staff, and students on main and regional campuses and on home computers. Licenses purchased with OSU funds remain with OSU. To purchase, call 800-782-8272 or go to the StataCorp site. - Free SYSTAT (release 13) Windows 2000/XP/Vista/7
Missing Data
- Clinical Trial Design and Informed Consent Language to Avoid Missing Data Bias
Missing data reduces the benefit provided by randomization, and may lead to biased conclusions. It can arise from a variety of reasons, including the inability or unwillingness of participants to meet appointments for evaluation. In addition, inadvertent loss of data occurs when participants, who discontinue treatment, are no longer followed.
Miscellaneous Resources
These are a few concepts that need special attention when interpreting the statistical results:
- The Prism Guide to Interpreting Statistical Results
Analyzing Data with GraphPad Prism - Course notes on confidence intervals
Department of Statistics, Yale University - Balancing statistical and clinical significance in evaluating treatment effects
W-C Leung, Postgrad Med J 2001;77:201–204
Glossary
- Glossary of key terms in Statistics
Writing at Colorado State University