Follow the procedure to select the appropriate statistical method

Home

Method list

Case study

Discussion forum

Click the statement that fits your case description:

• Your question is to investigate features of a single variable;
  • The variable is continuous, then
    • begin from one sample t test.
  • The variable is categorical, then
    • begin from Goodness-of-fit test.
• Your question is to investigate effect of (independent) variable(s) on other (dependent) variable(s);
  • There is one dependent variable and one independent variable;
    • The dependent variable is continuous;
      • The independent variable is categorical;
        • The independent variables has two categories (e.g., female/male);
          • The samples are independent, then
            • begin from two sample t-test ;
            • or use Wilcoxon rank sum test if you know data is not normally distributed
          • The samples are dependent, then
            • begin from paired t-test .
            • or use Wilcoxon sign rank test if you know data is not normally distributed.
        • The independent variables has three or more categories (e.g., black/red/white);
          • The samples are independent, then
            • begin from ANOVA .
            • or use Kruskal Wallis test if you know data is not normally distributed.
            • or use Welch test if you know data has heterogeneity variance.
          • The samples are dependent, then
            • suggest methods that consider repeated measurement.
      • The independent variable is continuous, then
        • begin from linear regression .
    • The dependent variable is categorical;
      • The independent variable is categorical;
        • The samples are independent(such as female/male), then
          • begin from contingency test .
          • or use Fisher's exact test , if you know the sample size is relatively small.
        • The samples are dependent (e.g, wife/husband), then
          • begin from McNemar's test .
      • The independent variable is continuous, then
        • begin from logit regression.
  • There is one dependent variable and multiple independent variables;
    • The dependent variable is continuous;
      • The independent variables are categorical, then
        • begin from N-way Analysis of Variance (ANOVA)
      • The independent variables are continuous, then
        • begin from linear regression .
    • The dependent variable is categorical, then
      • begin from logit regression.