🚁 Experiment design and data analysis using R

Prerequisite: download RStudio or use Colab

Check the assumption of normality of our data

• Shapiro-Wilk Test:

normality_check <- shapiro.test(mydata$myDependentVariable)

if (normality_check$p.value > 0.05){

  print("The data comes from a population that is normally distributed. Do parametric statistical analysis")

} else {

  print("The data is not normally distributed. Do non-parametric statistical analysis")

} 

Compare two conditions

• Between-subject design: two conditions are independent

  • Parametric statistical analysis: Independent t-test (Two-sample t-test)

  • Non-parametric statistical analysis: Mann-Whitney U test

• Within-subject design: two conditions are dependent (e.g., pre-test v.s. post-test)

  • Parametric statistical analysis: Paired t-test

  • Non-parametric statistical analysis: Wilcoxon signed-rank test

Compare more than three conditions

• Between-subject design: three (or more) conditions are independent

  • Single factor:

    • Parametric statistical analysis: One-way ANOVA

    • Non-parametric statistical analysis: Kruskal-Wallis test

  • Two factors:

    • Parametric statistical analysis: Two-way ANOVA

    • Non-parametric statistical analysis: ART + two-way ANOVA

• Within-subject design: three (or more) conditions are dependent (e.g., pre-test v.s. mid-test v.s. post-test)

  • Single factor:

    • Parametric statistical analysis: Repeated measures ANOVA

    • Non-parametric statistical analysis: Friedman test

  • Two factors:

    • Parametric statistical analysis: Two-way repeated measures ANOVA

    • Non-parametric statistical analysis: ART + two-way repeated measures ANOVA

• Mixed factorial design: one factor is dependent, and another factor is independent

  • • Parametric statistical analysis: Mixed ANOVA

    • Non-parametric statistical analysis: ART + mixed ANOVA

🎠 Plot the results

• Boxplot with symbols showing significant differences. [Code]

⛵️ Self-learning and debugging

Try googling different combinations of keywords: 

E.g.: R t-test example; R [error message] stackoverflow; R colab install packages, etc.