Sample size calculations: How many observations to take?

At the basis of most experimental designs lies the question, ‘How many observations or experimental units should I include in my experiment’. Too few and the results could be disappointing, too many and it becomes infeasible or too expensive.

In this project, you will try to answer the sample size question by use of simulations for multivariate data. By studying small pilot data sets, you will get an idea of the variation you can expect in the large scale experiment. By simulating data sets, based on the pilot data, of varying sample sizes, you will try to find the minimal data set for which you still can correctly classify a set samples or create an optimal prediction model for sensory traits.


An interest in statistical methods, programming and life sciences (in random order). Familiarity with programming would help (e.g. matlab, r, python).