Statistical Methods Research Topics

A literature review of some statistical method in the style of a TREE (Trends in Ecology and Evolution) paper. The review should be critical not just descriptive. That is, enter the debate. Is post hoc power analysis a useful tool or not? When should we use least squares and when should we use a reduced major axis regression? Should we ever analyze ratios? Why do so many people use principal components analysis to measure differences between groups? How do we compare body size-correlated measures (egg size for example) among populations if these populations differ in body size? Are resampling methods always better than parametric methods? Should we abandon the Bonferroni world-view? Should we even bother with frequentist (P-value) statistics?

If you want to do something original, pour through recent issues of some journal in some field and count how people misinterpret p-values (e.g. "these two groups are really different (P=0.000000000000001)"). Or, how often do people use a Model I regression when they should have used model II. Or, how often could people have used a directional test but didn't? Or, how many people use PCA when they should have used DFA? I would love to a paper that included this sort of original [library] data!

Target dates

Please e-mail me about ideas and I want you to get started by the week after break. That is, on 7 March, I want an summary of your paper with at least 5 references. The paper will be due May 9, which is the monday of exam weeks. No length limit. Any fewer than 5 references will almost certainly not be sufficient research on your part.

Topics

  1. Power analysis
  2. Pseudoreplication
  3. Directional tests
  4. Bonferroni - pros and cons
  5. Meta-analysis
  6. Resampling statistics
  7. Model II regression
  8. Multiple regression vs. path analysis
  9. analysis of ratios
  10. size and shape
  11. phylogenetic regression
  12. spatial statistics
  13. multivariate
    1. correspondence analysis
    2. principal components analysis vs. canonical variates analysis
    3. cluster analysis
  14. model comparison, model averaging