Measures of Centrality
Context: FIT1043_MOC, FIT2086_MOC · the “typical value” of a sample · the first of the descriptive statistics · numerically interpreting numeric data FIT2086 framing: a statistic is any function of a sample; the mean , median and mode are the simplest — later reused as estimators of population parameters (see Statistical Modelling and Inference).
Quick Revision
- 🎯 Objective: summarise a sample’s centre ➔ mean, median, mode.
- 📦 Core Components: mean (all values) | median (middle value) | mode (most frequent).
- ⚡ Key Constraint: the mean is sensitive to every value (outliers drag it); the median is resistant — their gap reveals skew.
📝 How It Works
1. What is a Statistic?
- Descriptive statistics ➔ numerically interpret key features of a dataset; usually lose information but gain comprehension (contrast inferential statistics).
- Statistic ➔ for a sample , any function of the data.
2. The Three Measures
- Mean ➔ arithmetic average ; uses all values.
- Median ➔ = value with 50% of samples below it; sort and take the middle.
- Mode ➔ the most frequently occurring value.
3. Mean vs Median (robustness)
- Mean ➔ any change to any value changes it; one huge value can move it arbitrarily.
- Median ➔ uses at most two middle values ➔ resistant to changes away from the middle.
- Worked ➔ ; .
⚖️ Core Decision Matrix
| Distribution | Mean vs Median | Tail |
|---|---|---|
| Symmetric | mean median | balanced |
| Positively skewed | mean median | long right tail |
| Negatively skewed | mean median | long left tail |
When It Flips: the sign of (mean − median) diagnoses skew — the mean chases the long tail while the median stays put; equal ⇒ symmetric.
🧠 Active Recall
Why does changing one value from 5 to 50 move the mean from 3 to 12 but leave the median at 3?
- Hint: All-values vs middle-value.
Answer
- Short answer: The mean sums all values, so a large outlier inflates it; the median depends only on the middle position, unchanged when a non-middle value grows.
- Why: Resistance ➔ median uses at most two central values, so it’s robust to outliers; the mean is not.
How do you read skew from the mean and median?
- Hint: Mean chases the tail.
Answer
- Short answer: mean > median ⇒ positively skewed (right tail); mean < median ⇒ negatively skewed (left tail); mean ≈ median ⇒ symmetric.
- Why: Tail pull ➔ extreme values in the tail drag the mean toward that tail.