Analysis of Variance (ANOVA)

Comparing the means of 3 or more groups simultaneously. Why not just do multiple t-tests? Because that increases the risk of Type I error!

The Concept

ANOVA splits the total variation in the data into two parts:

  • Between-Group Variation: Differences caused by the treatment (e.g., different drugs).
  • Within-Group Variation: Random noise or error.
$$ F = \frac{\text{Variance Between Groups}}{\text{Variance Within Groups}} $$

If $F$ is large, the treatment has a significant effect.

One-Way ANOVA

Used when there is only one factor (independent variable). Example: Effect of 3 different fertilizers on crop yield.

ANOVA Table

Source Sum of Squares (SS) df Mean Square (MS) F
Between SSB $k-1$ $MSB = SSB/df$ $MSB/MSW$
Within SSW $N-k$ $MSW = SSW/df$
Total SST $N-1$

Two-Way ANOVA

Used when there are two factors. Example: Effect of Fertilizer AND Watering Frequency on crop yield.

It allows us to test for Interaction Effects (e.g., maybe Fertilizer A works best only with high watering).

Test Yourself

Q1: If the F-statistic is close to 1, what does it mean?

  • No significant difference between groups
  • Significant difference exists
  • Calculation error