Sample size and significance – somewhere between statistical power and judgment prostration

When performing scientific research we are so “embraced” to use the tool of inductive logic in our reasoning that we often express more generalized opinions on the population of interest based on relatively small sample(s) of a general population. What we take care about in such situations is that chosen segments are representative for a whole set of elements in the general population. To cope with such a demand we always want to know how large our selected subpopulation should be to enable us to detect the experimental effect of interest not only at a certain level of significance, but also with the highest possible power of statistical reasoning. Thus, when designing our experiment, we have to compromise between a sample size not too small to ensure that our sample is sufficiently representative, and not too large to benefit from the sampling procedure at all. The tools for the estimation of minimum required sample size and the analysis of power, which help us to make quick decisions on how to compromise reasonably between significance, statistical power and sample size, are discussed in this paper.