Statistics

We’re happy to do any of the following kinds of analysis:

Regression and ANOVA, including specialized variations such as logistic regression (to estimate how a probability varies, using success-failure data) and Poisson regression (to estimate how a rate varies, using event count data)
Nonparametric analysis, from simple Kruskal-Wallis tests and finding tolerance intervals (bounds that will contain p% of your data q% of the time), to performing the nonparametric equivalent of 2- or 3-way ANOVA using the Aligned Rank Transform
Explicitly writing out the model form and uncertainty structure for a complicated model, and directly applying maximum likelihood estimation to it. This last is a particular specialty of ours. Rather than shoehorning your data to fit into a technique that you can find in your freshman statistics book, we construct a method custom-designed for your particular problem.
Design of survey instruments and sampling schemes, and analysis of the responses you collect to your survey. (We don’t offer to administer the survey on your behalf.)

If you have something else in mind, go ahead and ask us; we'll let you know if it's something we can do for you or not.

Consulting a statistician during the experimental design phase almost always saves you time and money, compared to collecting your data set before contacting us. We can help you design an efficient data collection scheme, and ensure that your data will really be suitable for answering your research questions.

This page last edited 30.10.17