Choosing Analysis Methods by Variable Types
You'll learn: How to pick the right bivariate technique by identifying whether you're comparing numeric-numeric, categorical-categorical, or mixed variable pairs.
Why Variable Types Matter
When you explored univariate analysis, you learned that numeric and categorical variables require different visualization and summary techniques. The same principle applies when examining relationships between two variables—the combination of variable types determines which methods make sense.
Think of it like choosing the right tool for a job: you wouldn't use a hammer to tighten a screw. Similarly, you can't calculate a correlation coefficient between two categorical variables (like "color" and "shape"), but it works perfectly for numeric pairs (like "height" and "weight").
The Three Main Scenarios
Numeric × Numeric: When both variables are continuous or discrete numbers, you want to assess linear relationships, strength, and direction. Scatter plots and correlation coefficients are your primary tools.
Categorical × Categorical: When comparing two categorical variables (like "education level" and "income bracket"), you examine how the distribution of one category changes across another. Cross-tabulation (contingency tables) and grouped bar charts help here.
Mixed (Numeric × Categorical): When one variable is numeric and the other categorical (like "salary" across "department"), you're comparing distributions of the numeric variable across groups. Box plots by group and comparing group means are effective approaches.
Making the Decision
Before diving into analysis, always ask: "What type is each variable?" This single question guides your entire analytical approach and prevents methodological mistakes.
Key Takeaway: Match your bivariate analysis method to your variable type combination—numeric-numeric pairs use correlation and scatter plots, categorical-categorical pairs use cross-tabs, and mixed pairs compare numeric distributions across categories.