Please enable JavaScript.
Coggle requires JavaScript to display documents.
Statistical Analysis, Cazetta, References: Data Science Academy. (n.d…
Statistical Analysis
Start
Is your sample size ≥ 10 per group?
Yes
Is your dependent variable quantitative?
Yes
Description of one group
Ratio and Interval Data
Normal distribution
Mean, standard deviation
Non-normal distribution
Median, interquartile range
Ordinal data
Median, interquartile range
Comparison of one group to hypothetical value
Ratio and Interval Data
Normal distribution
One-way t-test
1 more item...
Non-normal distribution
Wilcoxon test
1 more item...
Ordinal data
Wilcoxon test
Paired groups, non-normal data
Comparison of two groups
Unpaired groups
Ordinal data
Mann-Whitney test
Ratio and Interval Data
Non-normal distribution
1 more item...
Normal distribution
2 more items...
Paired groups
Ratio and Interval Data
Normal distribution
1 more item...
Non-normal distribution
1 more item...
Ordinal data
Wilcoxon test
Comparison of three or more groups
Unmatched groups
Ratio and Interval data
Normal distribution
1 more item...
Non-normal distribution
1 more item...
Ordinal data
Kruskal–Wallis's test
1 more item...
Matched groups
Ratio and Interval data
Normal distribution
1 more item...
Non-normal distribution
1 more item...
Ordinal data
Friedman's test
1 more item...
Measure association between two varaibles
Ratio and Interval data
Normal distribution
Pearson's test
1 more item...
Non-normal distribution
Spearman's correlation
1 more item...
Ordinal data
Spearman's correlation
Prediction
From another measured variable
Ratio and Interval data
Normal distribution
1 more item...
Non-normal distribution
1 more item...
Ordinal data
Non-parametric regression
1 more item...
From several measured or binomial variables
Ratio and Interval data
Multiple linear or nonlinear regression
1 more item...
Ordinal data
Multiple logistic regression
1 more item...
No
Description of one group
Proportion
Comparison of one group to hypothetical value
Chi-squared
Testing association between two categorical variables
Comparison of two groups
Unpaired groups
Fisher's test (chi-squared for large samples)
Like Chi-squared, but for very small categorical datasets (2×2 tables)
Paired groups
McNemar's test
Tests paired nominal data, e.g. same group before and after (yes/no responses)
Comparison of three or more groups
Unmatched groups
Chi-squared
Testing association between two categorical variables
Matched groups
Cochran's Q test
Compare three or more related groups with binary outcomes (e.g., success/fail across 3+ conditions) — non-parametric and used for repeated yes/no data
Measure association between two varaibles
Contigency coefficients
Measures strength of association in contingency tables
Prediction
From another measured variable
Simple logistic regression
Predicts a binary outcome (e.g., germinates: yes/no) from one predictor
From several measured or binomial variables
Multiple logistic regression
Predicts a binary outcome using multiple predictors
No
Is your sample size between 5–9?
Yes
Is your data ordinal or non-normal?
Use non-parametric test only
Compare 3+ groups (non-normal)
Kruskal–Wallis test
1 more item...
Related measures (e.g., repeated trials)
Wilcoxon or Friedman
1 more item...
Non-linear or ranked correlations
Spearman
1 more item...
Compare 2 groups (non-normal)
Mann–Whitney U test
1 more item...
No
Sample too small — increase sample size
Cazetta
References:
Data Science Academy. (n.d.). Análise Estatística para Data Science com Linguagem R [Online course].
https://www.datascienceacademy.com.br/
Figure adapted from Pani, A. K., Mishra, P., & Biswal, M. R. (2020). A comprehensive guide for statistical analysis in research.
https://www.researchgate.net/publication/345573209
Statistical concepts were supported and organized with the help of OpenAI’s ChatGPT.