University of Twente Student Theses


Graphical exploration of statistical assumptions : can we?

Brandl, L. (2016) Graphical exploration of statistical assumptions : can we?

[img] PDF
Abstract:In statistical inference, failure to control for violations of statistical assumptions increases the risk of committing a type I (rejecting a true null hypothesis) or type II error (failure to reject an untrue null hypothesis). Graphical exploration of statistical assumptions is advocated. The objective of the current study was to confirm the usefulness and superiority of statistical graphics in data exploration. A complete within-subject design was employed, exposing participants to 100 simulated dot-histograms and 100 box-jitter plots. Participants' ability to visually detect violations of the normality assumption and homoscedasticity was assessed. Results revealed that participants were not able to validly detect violations of statistical assumptions. Exploratory data analysis and a general linear mixed-model in form of a logistic regression further uncovered that participants did not inform their choices by objective criteria. However, our results are ambiguous. Participants varied in their baseline tendency to reject stimuli and in how much they were influenced by objective criteria. Stimuli varied in which kind of response they provoked, even after objective criteria were taken into account. We conclude that the scientific community is not ready for a methodological shift from conventional statistical techniques to graphical data exploration. A temporary joint usage of statistical tests and graphics is advocated while further research investigates whether our results can be replicated with subjects of higher statistical proficiency. Drastic changes to the psychological curriculum and empirical research into the effectiveness of competing statistical graphics shall prepare the community for the supersession of conventional statistical techniques.
Item Type:Essay (Bachelor)
Faculty:BMS: Behavioural, Management and Social Sciences
Subject:70 social sciences in general, 77 psychology
Programme:Psychology BSc (56604)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page