Welcome to Skewness: Understanding the Shape of Data
Welcome to your study notes on Skewness! In Statistics, finding averages like the mean or median gives us a great summary of the centre of a dataset, but it doesn't tell us the whole story. We also need to know how the data is shaped and spread out.
Don't worry if this sounds complicated at first. By the end of this guide, you will easily be able to look at a chart or a set of averages and instantly describe its skewness with total confidence!
1. What is Skewness?
Skewness is simply a measure of the asymmetry (or lack of symmetry) in a frequency distribution. In everyday language, it tells us whether the data is nicely balanced in the middle or stretched out more towards one side.
Symmetrical Distributions (Zero Skew)
When data is symmetrical, it is evenly balanced around the central point. If you were to draw a vertical line straight down the middle, the left half would look like a mirror image of the right half.
For a symmetrical distribution:
• The Averages: The three main averages are all roughly equal: \( \text{Mean} \approx \text{Median} \approx \text{Mode} \)
• The Box Plot: The median line (\( Q_2 \)) sits right in the middle of the box, meaning the distance between the lower quartile and the median is approximately equal to the distance between the median and the upper quartile: \( Q_2 - Q_1 \approx Q_3 - Q_2 \)
• The Whiskers: Both whiskers on a box plot are approximately equal in length.
Key Takeaway: A symmetrical distribution is perfectly balanced with no long tail on either side.
2. The Two Types of Skewness
When data is not balanced, it leans or stretches towards one side. There are two main types of skewness assessed in CCEA GCSE Statistics:
A. Positive Skew (Right-Skewed)
In a positively skewed distribution, the bulk of the data is clustered at the lower values (on the left), while a few unusually high values stretch out a long tail to the right.
Visual Shape: A steep peak on the left with a long tail pointing towards higher positive values on the right.
Relationship of the Averages:
Because extreme high values pull the mean upwards, the averages follow this strict order:
\( \text{Mode} < \text{Median} < \text{Mean} \)
Box Plot Features:
• The right whisker is noticeably longer than the left whisker.
• The median is closer to the lower quartile (\( Q_1 \)) than to the upper quartile (\( Q_3 \)), which means: \( (Q_3 - Q_2) > (Q_2 - Q_1) \)
Real-World Example: Household incomes. Most people earn typical, modest salaries (clustering at the lower end), but a small number of billionaires stretch the tail far to the right, pulling the mean income much higher than the median.
B. Negative Skew (Left-Skewed)
In a negatively skewed distribution, the bulk of the data is clustered at the higher values (on the right), while a few unusually low values stretch out a long tail to the left.
Visual Shape: A long tail pointing towards lower values on the left, with a steep peak on the right.
Relationship of the Averages:
Because extreme low values drag the mean downwards, the averages follow this strict order:
\( \text{Mean} < \text{Median} < \text{Mode} \)
Box Plot Features:
• The left whisker is noticeably longer than the right whisker.
• The median is closer to the upper quartile (\( Q_3 \)) than to the lower quartile (\( Q_1 \)), which means: \( (Q_2 - Q_1) > (Q_3 - Q_2) \)
Real-World Example: Exam scores on a very straightforward test. Most students score high marks (clustering near the top), but a few very low marks stretch the tail out to the left.
Memory Trick: "The tail tells the tale!" Always look at the direction the tail is pointing, not the hump. If the tail stretches right towards higher numbers, it is positive. If it stretches left towards lower numbers, it is negative.
3. Identifying Skewness from Statistical Charts
In your CCEA examination, you will often be asked to determine the skewness of a distribution directly from charts and diagrams. Here is how to inspect each type:
1. Histograms & Frequency Polygons
• Look at the slopes on either side of the highest bar or peak.
• If the frequency bars trail off gently towards the right, the distribution has positive skew.
• If the frequency bars trail off gently towards the left, the distribution has negative skew.
2. Stem-and-Leaf Diagrams
• Look at the lengths of the rows (leaves).
• If the longest rows are at the top (smaller values) and taper off towards the bottom (larger values), the diagram shows positive skew.
• If the longest rows are at the bottom (larger values) and taper off towards the top (smaller values), the diagram shows negative skew.
3. Box-and-Whisker Plots (Box Plots)
To check skewness on a box plot, use a two-step check:
1. Compare the whiskers: Is one whisker much longer than the other?
2. Compare the sections of the box: Compare the distance from the lower quartile to the median, \( (Q_2 - Q_1) \), with the distance from the median to the upper quartile, \( (Q_3 - Q_2) \).
• If \( (Q_3 - Q_2) > (Q_2 - Q_1) \) and the right whisker is longer \(\implies\) Positive Skew
• If \( (Q_2 - Q_1) > (Q_3 - Q_2) \) and the left whisker is longer \(\implies\) Negative Skew
• If \( (Q_3 - Q_2) \approx (Q_2 - Q_1) \) and whiskers are equal \(\implies\) Symmetrical
4. Calculation Rules & Summary Formulas
When given summary statistics rather than diagrams, you can use these simple calculation checks:
Quartile Test
Calculate the two halves of the interquartile range:
• Step 1: Calculate the upper gap: \( \text{Upper Gap} = Q_3 - Q_2 \)
• Step 2: Calculate the lower gap: \( \text{Lower Gap} = Q_2 - Q_1 \)
• If \( \text{Upper Gap} > \text{Lower Gap} \), the data is positively skewed.
• If \( \text{Lower Gap} > \text{Upper Gap} \), the data is negatively skewed.
• If \( \text{Upper Gap} = \text{Lower Gap} \), the data is symmetrical.
Averages Difference Test
Compare the position of the mean relative to the median or mode:
• \( \text{Mean} - \text{Median} > 0 \) (or \( \text{Mean} - \text{Mode} > 0 \)) \(\implies\) Positive Skew
• \( \text{Mean} - \text{Median} < 0 \) (or \( \text{Mean} - \text{Mode} < 0 \)) \(\implies\) Negative Skew
• \( \text{Mean} - \text{Median} \approx 0 \) \(\implies\) Symmetrical
5. Examiner Tips & Common Pitfalls to Avoid
Examiners frequently report the same few mistakes on skewness questions in CCEA exams. Make sure you don't lose easy marks!
Pitfall 1: Confusing the "Hump" with the "Tail"
The most common mistake students make is looking at where the main cluster (the peak) is located. If the peak is on the left, students mistakenly write "negative skew".
Remember: Skewness is named after the tail, not the peak! A peak on the left means the tail points to the right \(\implies\) Positive Skew.
Pitfall 2: Forgetting Context in Comparison Questions
When comparing two distributions (e.g., test scores before and after tutoring), don't just state "the skewness changed from positive to negative". Always link your observation to what it actually means in context using averages and spread:
• Mention the change in the average (e.g., "The median increased from 42 to 68...")
• Mention the change in spread (e.g., "...and the IQR decreased, showing marks became more consistent.")
Pitfall 3: Whisker Outlier Confusion
An extreme outlier can cause one whisker on a box plot to stretch very far, giving an appearance of skewness even if the middle 50% of the data (the box) is balanced. Always inspect both the box symmetry (\( Q_3 - Q_2 \) vs \( Q_2 - Q_1 \)) and the whisker lengths to give a complete answer.
Quick Summary Checklist
Before moving on, make sure you can answer these three questions:
1. If \( \text{Mean} = 55 \), \( \text{Median} = 48 \), and \( \text{Mode} = 42 \), what is the skewness?
Answer: Positive Skew (because \( \text{Mode} < \text{Median} < \text{Mean} \)).
2. If \( Q_1 = 12 \), \( Q_2 = 20 \), and \( Q_3 = 24 \), what is the skewness?
Answer: Negative Skew (because \( Q_2 - Q_1 = 8 \), which is larger than \( Q_3 - Q_2 = 4 \)).
3. Where does the tail point in a negatively skewed distribution?
Answer: Towards the left (lower values).