CUERP Faculty Collaboration
Statistics and Biology Faculty Collaborate to Inform of New Statistical Methods
Dr. Timothy O'Brien, faculty member of Mathematics and Statistics and Dr. Martin Berg with the Department of Biology collaborated to highlight the importance of using appropriate analytical techniques to increase power in statistical modeling and testing. Their article "Getting the Most from Data - Maximizing Information and Power by Using Appropriate and Modern Statistical Methods" to be published in Journal of Data Science updates practitioners of some recent developments of techniques that are useful for biometric and biomedical research, and modeling.
The authors demonstrate with five examples the advantages of statistical modeling, of examining original data instead of collapsing or pooling them, of incorporating pairing into data analysis, and of choosing the statistical technique that maximized power and whose assumptions match the analysis. Specifically, these include:
- Misapplying the usual generic chi-square test; choosing a good statistical model.
- Collapsing data; testing for goodness of fit in Logistic regression.
- Accounting for the correlations inherent in paired count data.
- Analyzing ANOVA data with unequal variances using the Likelihood Ratio (LR) test.
- Modifying Logistic Regression for Interval-Censored Bioassay Data.
Overall, applied statistical analysis is a field that is in a continuous state of change and improvement because of better understanding, increased computing power and improved statistical packages.