Please enable JavaScript.
Coggle requires JavaScript to display documents.
What is your main research question, Normality Test: Always run a…
What is your main research question
PURPOSE OF ANALYSIS
What do you want to do?
Association / Correlation
Both variables are Continuous
Pearson (normal continuous)
Spearman (ordinal / non-normal)
Both variables are Categorical:
Chi-Square Test of Independence
Prediction / Adjustment
OUTCOME TYPE
continuous
Linear regression
count
Poisson / Neg. binomial
binary
Logistic regression
Compare Groups
OUTCOME TYPE
Continuous
How many groups?
2 Groups
INDEPENDENCE CHECK
Independent
Normal distribution → Independent t-test
Non-normal → Mann–Whitney U
Paired
Normal → Paired t-test
Non-normal → Wilcoxon signed-rank
≥3 Groups
INDEPENDENCE CHECK
Repeated
Normal → Repeated-measures ANOVA
Non-normal →Friedman test
1 more item...
Independent
Normal → One-way ANOVA
Non-normal →Kruskal–Wallis
1 more item...
1 Group
One-sample t-test / Wilcoxon
Categorical
INDEPENDENCE CHECK
Paired binary data
McNemar test
Independent groups
• Expected counts ≥5 → Chi-square
• Small samples → Fisher’s exact
Time to Event
Unadjusted comparison
Kaplan–Meier + Log-rank test
Adjusted analysis
Cox proportional hazards regression
Agreement
OUTCOME TYPE
Categorical ratings
Cohen’s kappa
Continuous measurements
Intraclass correlation (ICC)
Normality Test: Always run a Normality Test first
Shapiro-Wilk or Kolmogorov-Smirnov
If p<0.05, your data is not normal, and you should likely use a non-parametric test.
Outcome type → Number of groups → Independence → Distribution → Test