Original language | English |
---|---|
Title of host publication | International Encyclopedia of Transportation |
Editors | Roger Vickerman |
Publisher | Elsevier |
Pages | 87-95 |
Volume | 7 |
ISBN (Electronic) | 978-0-08-102672-4 |
DOIs | |
Publication status | Published - 2021 |
Abstract
This article highlights four common pitfalls in the use of statistics in the area of traffic psychology. Through computer simulations of scenarios that are typical in the field, it is first shown that a statistically significant P-value does not prove that the effect is true, especially when the effect is surprising and the P-value barely significant. Second, we show that “everything is correlated”, a phenomenon which has important ramifications for significance testing. Third, we explain the perils of two-stage testing and data peeking. Finally, we explain that the violation of independence can easily lead to false positives.
Bibliographical note
Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-careOtherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.
Keywords
- Bias
- Common method variance
- Correlation coefficient
- False positives
- Hypothesis testing
- Outliers
- Power
- Questionable research practices
- Response style
- Sample size
- Statistical significance
- t-test
- Test assumptions
- Type I errors
Fingerprint
Dive into the research topics of 'Pitfalls of Statistical Methods in Traffic Psychology'. Together they form a unique fingerprint.Datasets
-
Script that produces the table and figures for: Pitfalls of Statistical Methods in Traffic Psychology
de Winter, J. C. F. (Creator), TU Delft - 4TU.ResearchData, 19 Nov 2020
DOI: 10.4121/13141289
Dataset/Software: Software