Statistical Analyses

So you have your data and you need to know what to do with it. Are you testing a hypothesis or making a predictive model? How is your data distributed (Normal, Gamma, Binomial, Poisson, or something else)? Is the data homoscedastic, structured temporally, spatially, or hierarchically? Which variables should be included in your statistical model?

How do you deal with these and many more questions? Even non-parametric tests have several key assumptions that, sadly, too many people are unaware of.

Choice of statistics: I will review your objective, experimental design, and data to determine the best statistical methods. I will provide you with:

  • the simplest analysis that answers your question correctly
  • a justification to make sure you understand the choice
  • a list of diagnostics to be done, and how these will guide your analyses


Completed analysis: If you are not comfortable or do not have the time to perform the analysis yourself, I can do it. I will provide:

  • information about the choice of statistics as above
  • the statistical results with their interpretation in relation to your question or objective
  • figures that illustrate your results. These can follow a specific format (e.g. A journal)
  • descriptive statistics and graphs of your data


Example of statistical analyses

  • choosing statistical methods according to your data and objectives
  • verification of assumptions and diagnostics
  • standard parametric tests: t-tests, ANOVAs, Pearson’s correlation, least square regression, general linear models
  • standard nonparametric tests: Kruskal-Wallis, Mann-Whitney U, Spearman correlation, and rank based methods
  • robust/resampling statistics: bootstrap, jackknife, permutations
  • nonlinear models: curve fitting, quadratic, exponential, logistic…
  • generalized linear models: Normal, Poisson (counts), Binomial (proportions), Gamma and other distributions, generalized estimating equations
  • mixed models: fixed factor, random factor, correlation structure
  • survival analyses: Cox proportional hazard, time-dependent variables
  • phylogenetic regression: sequence alignment, tree building
  • model selection: information criteria, AIC, BIC, QIC, model averaging