Piotr Wolski
Medycyna Nowożytna, Tom 29 (2023) Zeszyt 1, 2023, s. 361-379
https://doi.org/10.4467/12311960MN.23.019.18460Piotr Wolski
Rocznik Kognitywistyczny, Tom 9, 2016, s. 27-35
https://doi.org/10.4467/20843895RK.16.003.5471
In the present essay, the first in a short cycle, the author reviews and comments on the problems students and researchers have with proper understanding of the basics of statistical inference. Those difficulties seem to be in part due to mixing of the opposing theoretical stances of Fisher and Neyman, reviewed shortly. The author believes that the inconsistent standards of statistical inference afflict the teaching of methodology particularly.
Piotr Wolski
Rocznik Kognitywistyczny, Tom 9, 2016, s. 59-70
https://doi.org/10.4467/20843895RK.16.006.6412Statistical significance II. Interpretive pitfalls
The second of the series of essays on the problems of significance testing in psychological research focuses on inconsistencies of the logic of these tests and resulting problems with interpretation. The limits of their practical usability have been discussed, and reasons of their failure with a priori unlikely null-hypotheses explained. Misleading connotations of the term “statistical significance” have been stressed, that obscure the true meaning of statistical significance and promote bad practices, including overestimation of significance, and neglecting the problem of effect size.
Piotr Wolski
Rocznik Kognitywistyczny, Tom 9, 2016, s. 71-85
https://doi.org/10.4467/20843895RK.16.007.6413Statistical significance III. From ritual to statistical thinking
One of the more prominent problems of significance testing is ritualisation of their practical use and interpretation. In the present, third part of the series, reasons and manifestations of that rigidity have been discussed, and an alternative, sometimes labeled “statistical thinking”, presented. Matching a statistical significance testing scenario to the needs of the specific research program constitutes a part of statistical thinking. Some typical scenarios have been described, with the intent of showing how the same statistical tool, depending on it’s assumptions, can have differing use in research.