Introduction to Spearman’s Rank Correlation Coefficient

In the previous chapters, we looked at scatter diagrams to see if two variables were related. Sometimes, a scatter diagram shows a general trend, but the points don't form a perfect straight line. This is where Spearman’s rank correlation coefficient (often written as \(r_s\)) comes in!

Think of it as a way to measure how well the order (or rank) of two sets of data matches up. For example, if the person who came 1st in a Maths test also came 1st in a Science test, and the person who came last in Maths also came last in Science, there is a very strong relationship between the ranks.

Important Note: If you are sitting the Foundation Tier, you only need to know how to interpret a given value. If you are sitting the Higher Tier, you also need to know how to calculate it using a formula.

What is "Ranking" Data?

Before we can use Spearman’s rank, we have to turn raw data into ranks. Ranking simply means putting the data in order from smallest to largest (or largest to smallest).

Example: If three students scored 45, 80, and 62 in a test:
- 45 is the smallest, so it gets Rank 1.
- 62 is the next smallest, so it gets Rank 2.
- 80 is the largest, so it gets Rank 3.

Quick Tip: In your exam, you won't be asked to deal with "tied ranks" (where two numbers are the same). Every value in your data set will be unique!

Interpreting the Coefficient (Foundation and Higher)

The result of a Spearman’s rank calculation is always a single number between \(-1\) and \(+1\). Here is what that number tells you:

  • \(+1\): Perfect positive correlation. The ranks match exactly (1st stays 1st, 2nd stays 2nd, etc.).
  • Close to \(+1\) (e.g., \(0.8\)): Strong positive correlation. As one rank increases, the other tends to increase.
  • \(0\): No correlation. The ranks are completely random compared to each other.
  • Close to \(-1\) (e.g., \(-0.7\)): Strong negative correlation. As one rank increases, the other tends to decrease.
  • \(-1\): Perfect negative correlation. The ranks are exactly reversed (1st becomes last, last becomes 1st).
Putting it into Context

In the exam, you must explain what the number means for the specific story in the question. If \(r_s = 0.9\) for "Hours studied" and "Exam mark," don't just say "strong positive." Say: "There is a strong positive correlation, meaning students who ranked higher in hours studied also tended to rank higher in their exam marks."

Calculating Spearman’s Rank (Higher Tier Only)

For the Higher Tier, you will be given the formula on your formula sheet. It looks a bit scary, but it’s actually like a recipe—just follow the steps!

\(r_s = 1 - \frac{6 \sum d^2}{n(n^2 - 1)}\)

What do the symbols mean?
  • \(n\) = The number of pairs of data you have.
  • \(d\) = The difference between the two ranks for each pair.
  • \(d^2\) = The difference squared.
  • \(\sum d^2\) = The sum of all the squared differences.
Step-by-Step Guide:

Step 1: Rank the first set of data (from 1 to \(n\)).
Step 2: Rank the second set of data (from 1 to \(n\)).
Step 3: Find the difference \(d\) for each pair by subtracting one rank from the other.
Step 4: Square each difference to get \(d^2\). (This turns any negative numbers into positives!)
Step 5: Add up all your \(d^2\) values to get \(\sum d^2\).
Step 6: Put your numbers into the formula and solve.

Example: If \(n = 5\) and \(\sum d^2 = 4\):
\(r_s = 1 - \frac{6 \times 4}{5(5^2 - 1)}\)
\(r_s = 1 - \frac{24}{5(24)}\)
\(r_s = 1 - \frac{24}{120}\)
\(r_s = 1 - 0.2 = 0.8\)

Spearman’s Rank vs. PMCC (Higher Tier Only)

You might also hear about the Pearson’s product moment correlation coefficient (PMCC). It is important to know the difference:

  • PMCC measures how close the points are to a straight line.
  • Spearman’s Rank measures the relationship between ranks. It is more useful if the relationship is a curve rather than a straight line, or if there are extreme outliers that might "mess up" the PMCC.

Common Mistakes to Avoid

  • Forgetting the "1 - ": Many students calculate the fraction part perfectly but forget to subtract it from 1 at the very end.
  • Squaring incorrectly: Remember, any number squared (even a negative one) must be positive. If you get a negative \(d^2\), something is wrong!
  • Mixing up \(n\): \(n\) is the number of pairs of data, not the total number of individual data points.
  • Ranking backwards: It doesn't matter if you rank smallest to largest or largest to smallest, but you must do it the same way for both variables!

Quick Review Box

  • Spearman’s Rank (\(r_s\)) measures correlation between ordered lists.
  • Values range from \(-1\) (perfect negative) to \(+1\) (perfect positive).
  • A value of \(0\) means no association.
  • Context matters: Always relate your answer back to the labels in the question!

Did you know? Spearman's rank is great for "subjective" data, like two judges ranking contestants in a talent show. You can't "calculate" beauty or talent, but you can definitely rank it!