Skewness and Kurtosis
Okay so before we even touch ANOVA properly, we need to quickly check what our data looks like. Not in a scary technical way, just in a “is this behaving normally or not?” kind of way.
This is where skewness and kurtosis come in. Think of them as quick vibe checks for your distribution.
Skewness (the lean)
Skewness tells you whether your data is balanced or leaning to one side. If it’s close to 0, everything is nice and symmetrical. If it’s positive, your data is stretched out to the right. If it’s negative, it’s stretched to the left.
Kurtosis (the shape)
Kurtosis tells you whether your data is super peaked and intense, or more flat and spread out. A value around 3 is what we expect for normal data. Higher means more extreme values hanging out in the tails.
How to actually read the numbers
| Measure | Value | What’s happening | What you say in your prac |
|---|---|---|---|
| Skewness | ≈ 0 | Balanced and symmetrical | Looks normal |
| Skewness | > +1 | Leaning heavily to the right | Not normally distributed |
| Kurtosis | ≈ 3 | Normal peak | All good |
| Kurtosis | > 3 | More peaked + extreme values | Possible normality violation |
Realistically: if your skewness is way above 1 and your kurtosis is clearly above 3, your data is not behaving nicely. It doesn’t mean you panic, just that you acknowledge it and interpret your results carefully.
