PhD Tutor Consultant

I can help you, even with a sample size as small as 3 for each experimental and control group or a total of 6 for independent samples; or 5 for paired comparisons! And wait, this will shock you: even if it was collected non randomly! There are nonparametric techniques which approach, or (given that they can operate on such small samples with less restrictions), EXCEED the power of traditional statistics which require much larger N such as t-tests or ANOVA. And they don’t even have to be “normally distributed:”to work: that's how powerful they are!

These non parametric techniques are very useful in enterprises such as rapid appraisals, and urban and regional planning, where you unintentionally violate some or even most of your parametric assumptions; or in biomedical drug trials where your samples, say those dying of terminal cancer exposed to new drugs or new procedures cannot be large and cannot be randomly sampled. If your concept of probability is old school (classical asymptotic), and are skeptical about this claim, hear it straight from a mathematician: read "A Powerful Technique for Small Sample Non Random Testing" below. Just click on the link below

When things get tough, simulation techniques such as Monte Carlo, and resampling techniques, such as bootstrap come to the rescue. I can also do meta-analysis for you, where we make use of previously used data for a purpose different from your research questions. Just relax and let me do the computing for you. When using nonparametric statistics, I can compute exact probabilities manually (with a calculator of course), without recourse to a software like SPSS or SAS. Don’t despair!