Summary:
The objective of the paper is to draw attention to incorrectness of variaous statistical approaches to research data.
We critically assess the situation when first the decision on the hypothesis to be proved is done and then data are
made to fit the hypothesis. We demonstrate that data cannot be satisfactorily interpreted without close attention to
the manner of their collection. Finnaly we criticize opportunistic data tortuning when researcher explores the data
until a significant result is found and then devises a biologically plausible hypothesis.
Key words:
statistics, fraud, data dredging, validity of data, multiple testing.
|