Introduction to Outliers and Comparing Data Sets
In Statistics, we don't just look at single numbers; we look at the stories data tells us. Sometimes, a data point tells a story that is completely different from the rest—we call these outliers. Other times, we need to look at two different groups (like test scores from two different classes) and decide which one performed "better." This chapter will teach you how to spot the "odd ones out" and how to compare different sets of data fairly and accurately.
1. Identifying Outliers: The "Odd Ones Out"
An outlier is a data value that is inconsistent with the rest of the data set. Think of it as a giant standing in a room full of toddlers. You can identify outliers in two ways: by inspection (looking) or by calculation (maths).
A. Identifying by Inspection
When you look at a Box and Whisker Plot or a Scatter Diagram, an outlier usually sticks out. On a box plot, they are often marked with a small "x" or a dot outside the whiskers. On a scatter diagram, an outlier is a point that lies far away from the "cloud" of other points or the line of best fit.
B. Identifying by Calculation
The Edexcel syllabus requires you to know two specific mathematical rules for defining an outlier. You must memorise these as they are not in the formula booklet!
Method 1: The Interquartile Range (IQR) Rule
This is the most common method used with box plots and medians. A value is an outlier if it is:
- Smaller than \(LQ - 1.5 \times IQR\)
- Larger than \(UQ + 1.5 \times IQR\)
(Note: \(LQ\) is the Lower Quartile, \(UQ\) is the Upper Quartile, and \(IQR = UQ - LQ\)).
Method 2: The Standard Deviation Rule
This method is usually used when the data follows a Normal Distribution. A value is an outlier if it lies more than 3 standard deviations away from the mean:
- Outside the range \(\mu \pm 3\sigma\)
(Where \(\mu\) is the mean and \(\sigma\) is the standard deviation).
C. Why do Outliers Happen?
When you find an outlier, you must determine its nature. Ask yourself: - Is it an error? Maybe someone typed \(100\) instead of \(10\), or a sensor malfunctioned. These should usually be removed or corrected. - Is it natural variation? Sometimes, a value is just extreme but perfectly valid (like a world-record athlete's height). These should be kept as they represent the true population.
Key Takeaway: Always check the context. If a human's height is recorded as \(5\) metres, it's an error. If it's \(2.2\) metres, it's a genuine extreme value.
2. Comparing Data Sets
When an exam question asks you to "compare two data sets," it is looking for two specific things: a comparison of location (averages) and a comparison of spread (consistency).
A. Measures of Location (The "Average")
You can use the Mean, Median, or Mode to talk about the "typical" value. - Mean (\(\bar{x}\)): Use this for "normal" data without extreme outliers. It uses every bit of data. - Median: Use this if there are extreme outliers, as it is "robust" (it doesn't get pulled away by one weird number). - Mode: Use this for non-numerical data (like favourite colours).
B. Measures of Spread (The "Consistency")
Spread tells you how much the data varies. - Standard Deviation (\(s\)) or Variance (\(s^2\)): Tells you how far, on average, data points are from the mean. A smaller standard deviation means the data is more consistent. - Interquartile Range (IQR): Focuses on the middle \(50\%\) of the data. Like the median, it is not affected by outliers. - Range: The simplest measure (\(Highest - Lowest\)), but it is very sensitive to outliers.
C. How to Write a Comparison (The Winning Formula)
To get full marks in an AO2/AO3 question, you must follow this structure: Context + Value + Direction.
Example: Comparing the heights of plants in Sunlight vs. Shade.
- Bad Answer: "The mean for sunlight is higher." (No context, no values).
- Good Answer: "The mean height of plants in sunlight (\(45cm\)) is greater than those in the shade (\(32cm\)), suggesting sunlight increases growth."
Quick Review: Comparison Checklist
1. Compare one measure of location (e.g., Median).
2. Compare one measure of spread (e.g., IQR).
3. Always use units and context from the question.
3. Important Considerations
The Impact of Outliers
Outliers "pull" certain statistics toward them. - The Mean and Standard Deviation are heavily affected by outliers. - The Median and IQR are resistant. - Memory Trick: The "M" in Mean stands for "Moves" (it moves easily!), while the Median stays "Middle-ground."
Calculator Usage
For Paper 1 and Paper 2, you are expected to find the Mean (\(\bar{x}\)), Standard Deviation (\(s\)), and Variance (\(s^2\)) directly from your calculator. - Ensure you use the \((n-1)\) divisor for sample variance, which is the standard for the 9ST0 specification. - Don't waste time doing these by hand unless the question specifically asks for a "show that" calculation!
4. Common Mistakes to Avoid
1. Forgetting the \(1.5\): Students often forget the multiplier in the outlier formula. It is \(1.5 \times IQR\), not just \(IQR\).
2. Comparing the wrong things: Don't compare the Mean of Group A to the Median of Group B. Compare Mean to Mean or Median to Median.
3. Ignoring Context: In Statistics (9ST0), "The spread is smaller" is a math fact. "The runners' times were more consistent in the second race" is a statistical conclusion. Always aim for the latter!
Key Takeaway: Statistics is about interpretation. If you calculate an outlier, always finish by saying what that outlier might represent in the real world.