Course contentsShow
← Data Science
Lesson 528 of 2,14514. Simple Linear RegressionFree lesson

Standardized vs Unstandardized Coefficients

Comparing raw coefficients to standardized beta coefficients and when each is more interpretable.

Standardized vs Unstandardized Coefficients

What you'll learn: You'll understand the difference between raw regression coefficients and standardized beta coefficients, and when each helps you interpret your model better.

The Core Difference

When you fit a simple linear regression, you get an unstandardized coefficient (the slope β₁) that tells you: "For every 1-unit increase in X, Y changes by β₁ units." This coefficient preserves the original units of your variables.

A standardized coefficient (often called beta or β*) answers a different question: "For every 1 standard deviation increase in X, Y changes by β* standard deviations." You calculate it by first converting both X and Y to z-scores (mean = 0, standard deviation = 1), then running the regression.

When to Use Each

Use unstandardized coefficients when:

  • You need real-world interpretation ("Each additional year of education increases salary by $5,000")
  • Your units are meaningful and familiar
  • You're making practical predictions

Use standardized coefficients when:

  • Comparing effects across variables with different units (e.g., "Does age or income affect happiness more?")
  • Your original units are arbitrary or hard to interpret
  • You want to know relative importance of predictors

The Relationship

Think of it like currency conversion. The unstandardized coefficient is like saying "This costs 100 yen." The standardized coefficient converts everything to a common currency—standard deviations—so you can compare apples to apples.

Key Takeaway: Unstandardized coefficients preserve original units and are great for practical interpretation; standardized coefficients put everything on a common scale, making them ideal for comparing the strength of different predictors.