- Correlation measures direction and strength of association; it does not prove causation.
- Positive association means variables tend to rise together; negative association means one rises as the other falls.
- A scatter diagram helps identify direction, curvature, clusters and outliers.
- Pearson r measures linear association and lies between −1 and +1 when defined.
- Pearson r is undefined if either variable has zero variance.
- Spearman rho measures association of ranks; the untied shortcut is 1 − 6Σd²/[n(n²−1)].
- For tied ranks, assign average ranks and calculate Pearson correlation on those ranks.
- Near-zero Pearson correlation does not exclude nonlinear dependence.
- Outliers, confounding and common trends can distort interpretation.
Use these notes alongside your current textbook. Check important formulas, worked answers and syllabus coverage with your teacher or official learning material.
Correlation , Class 11 Economics
Original supplementary guide for Statistics for Economics. All paired datasets are constructed for learning, not observed economic evidence.
Learning objectives
Identify positive, negative and zero linear correlation; read scatter diagrams; calculate and interpret Karl Pearson's coefficient; compute Spearman's rank correlation for untied and tied ranks; distinguish correlation from causation; and recognise nonlinear patterns and outliers.
1. Meaning and importance of correlation
Correlation describes the direction and strength of association between two variables. Economic applications include examining the relationship between income and expenditure, advertising and sales, or rainfall and agricultural output. A correlation coefficient summarises association in observed data; it does not by itself explain why the association exists.
2. Types of correlation
Positive correlation means larger values of one variable tend to occur with larger values of another; negative correlation means larger values tend to occur with smaller values. Zero Pearson correlation means no linear association, not necessarily no relationship. Perfect positive linear correlation has r = +1 and perfect negative linear correlation has r = −1. Correlation can also be linear or nonlinear; monotonic relationships consistently increase or decrease but need not be straight lines.
3. Scatter diagrams
A scatter diagram plots each observed pair (x,y) as a point. An upward-sloping cloud suggests positive association, a downward-sloping cloud suggests negative association, and a widely dispersed cloud without a linear pattern suggests weak linear correlation. Inspect curvature, clusters and extreme observations before computing a single summary statistic. Scatter plots do not demonstrate causal direction.
4. Karl Pearson's coefficient of correlation
For paired observations (xᵢ,yᵢ), Pearson's product-moment coefficient is r = Σ[(xᵢ − x̄)(yᵢ − ȳ)] / √{Σ(xᵢ − x̄)² × Σ(yᵢ − ȳ)²}. It is equivalently covariance divided by the product of standard deviations when numerator and denominator use consistent conventions.
A computational form is r = [nΣxy − (Σx)(Σy)] / √{[nΣx² − (Σx)²][nΣy² − (Σy)²]}. Pearson r is undefined if either variable has zero variance. The value lies from −1 to +1 when defined.
5. Pearson calculation: worked table
Constructed paired observations:
| x | y | x² | y² | xy | |
|---|---|---|---|---|---|
| 1 | 2 | 1 | 4 | 2 | |
| 2 | 3 | 4 | 9 | 6 | |
| 3 | 5 | 9 | 25 | 15 | |
| 4 | 4 | 16 | 16 | 16 | |
| Σ | 10 | 14 | 30 | 54 | 39 |
n = 4. Numerator = 4(39) − 10(14) = 16. Denominator = √{[4(30) − 100][4(54) − 196]} = √(20 × 20) = 20. Hence r = 16/20 = +0.8, indicating strong positive linear association in these four illustrative points. Do not infer a population-level effect from this tiny constructed dataset.
6. Pearson using deviations from means
For the same data, x̄ = 10/4 = 2.5 and ȳ = 14/4 = 3.5. Deviations dx are −1.5, −0.5, 0.5, 1.5 and dy are −1.5, −0.5, 1.5, 0.5. Then Σdxdy = 4, Σdx² = 5, Σdy² = 5, giving r = 4/√25 = 0.8. This cross-check agrees with the computational method.
7. Perfect positive and negative correlation
For pairs (1,2), (2,4), (3,6), all points lie on y = 2x, so r = +1. For pairs (1,6), (2,4), (3,2), all points lie on y = 8 − 2x, so r = −1. Perfect linear correlation is unusual in empirical economics and does not imply causality.
8. Interpreting the magnitude of r
The sign shows direction and the absolute value indicates strength of linear association. Values near +1 or −1 indicate a tight linear pattern; values near zero indicate weak linear association. Labels such as 'weak', 'moderate' and 'strong' have no universal cutoffs across disciplines. A coefficient is dimensionless and unaffected by changing measurement units through positive linear rescaling; reversing the direction of one variable changes the sign.
9. Spearman's rank correlation
Spearman's rho measures association between ranks and is suited to ordinal data and monotonic patterns. Rank both variables consistently (for example, smallest = rank 1), calculate each pair's rank difference d, and use the untied-rank formula ρ = 1 − 6Σd²/[n(n² − 1)]. This shortcut assumes no ties. For ties, assign average ranks and compute Pearson correlation on those ranks; simple untied formulas without adjustment can give incorrect results.
10. Spearman calculation: untied ranks
Illustrative rankings for four observations:
| Observation | Rank X | Rank Y | d | d² |
|---|---|---|---|---|
| A | 1 | 1 | 0 | 0 |
| B | 2 | 3 | −1 | 1 |
| C | 3 | 2 | 1 | 1 |
| D | 4 | 4 | 0 | 0 |
Σd² = 2, n = 4. ρ = 1 − 6(2)/[4(16 − 1)] = 1 − 12/60 = 0.8.
11. Tied ranks: worked example
Suppose X = (10, 10, 20, 30) and Y = (1, 2, 2, 4). Average ranks are RX = (1.5, 1.5, 3, 4) and RY = (1, 2.5, 2.5, 4). Their means are both 2.5. Deviations are RX − 2.5 = (−1, −1, 0.5, 1.5) and RY − 2.5 = (−1.5, 0, 0, 1.5). The cross-product sum is 3.75; squared-deviation sums are 4.5 and 4.5. Therefore rank Pearson correlation gives ρ = 3.75/4.5 ≈ 0.8333. Applying the no-ties shortcut directly would be inappropriate.
12. Pearson versus Spearman
| Aspect | Pearson r | Spearman ρ |
|---|---|---|
| Measures | Linear association | Monotonic rank association |
| Uses | Paired numerical observations | Ordinal ranks or ranked numerical data |
| Outlier sensitivity | Often substantial | Usually less sensitive to magnitudes, but not immune |
| Perfect association | Perfect straight-line pattern | Perfectly consistent rank ordering |
| Ties | Handled as numerical values | Use average ranks and a suitable tied-rank method |
13. Correlation is not causation
A high correlation does not prove that X causes Y. A third variable may influence both (confounding), the direction may run from Y to X (reverse causality), or the association may arise by chance. For example, ice-cream sales and heat-related illnesses may rise together because hot weather affects both. Causal conclusions require stronger study design and assumptions than correlation alone.
14. Nonlinearity, outliers and misleading associations
Pearson r can be near zero even when the variables have a clear curved relationship. A single extreme observation can dramatically change r. Spurious correlation can also arise from common trends over time or selective sampling. Always inspect a scatter plot, the number of observations, units, missing values and potential confounders before interpreting coefficients.
15. Additional worked examples
- Positive perfect linear: (1,2), (2,4), (3,6) → r = +1.
- Negative perfect linear: (1,6), (2,4), (3,2) → r = −1.
- Computational Pearson: n = 4, Σx = 10, Σy = 14, Σx² = 30, Σy² = 54, Σxy = 39 → r = 0.8.
- Spearman untied: n = 4, Σd² = 2 → ρ = 0.8.
- Zero Pearson with nonlinear pattern: x = −2, −1, 0, 1, 2 and y = 4, 1, 0, 1, 4 → r = 0 because Σxy = 0, although y = x².
- Rank ties: X = (10,10,20,30), Y = (1,2,2,4) → average-rank ρ ≈ 0.8333.
- Rescaling: changing X from rupees to paise multiplies X by 100 but leaves Pearson r unchanged.
16. Practice questions with answers
- Define correlation. A measure of the direction and strength of association between two variables.
- What does r = −0.9 suggest? Strong negative linear association in the observed data, not necessarily causation.
- What does r = 0 imply? No linear association as measured by Pearson r; a nonlinear relation may still exist.
- What are the bounds of Pearson r? −1 to +1, when defined.
- What is r for (1,2), (2,4), (3,6)? +1.
- What is r for (1,6), (2,4), (3,2)? −1.
- Find Spearman rho for n = 5 and Σd² = 4 with no ties. 1 − 24/[5(25 − 1)] = 0.8.
- Find Spearman rho for n = 4 and Σd² = 0 with no ties. +1.
- Why inspect a scatter plot? To identify direction, curvature, clusters and outliers.
- Why is the untied Spearman formula unsuitable for tied ranks? Ties change rank variability; use average ranks and correlation on ranks or a justified correction.
- Does r = +0.95 prove X causes Y? No; confounding or reverse causation may explain the association.
- What if all X observations are identical? Pearson r is undefined because X has zero variance.
17. Common mistakes and revision checklist
Do not claim correlation proves causation. Do not interpret r = 0 as absence of every possible relationship. Check that paired observations are correctly aligned. Do not mix covariance and standard-deviation denominators using inconsistent conventions. In Spearman calculations, use consistent rank direction and handle ties correctly. A correlation coefficient without a scatter plot, sample size or context may be misleading.
18. Formula sheet
| Method | Formula |
|---|---|
| Pearson deviation form | r = Σ[(x−x̄)(y−ȳ)] / √[Σ(x−x̄)² Σ(y−ȳ)²] |
| Pearson computational form | r = [nΣxy−ΣxΣy] / √{[nΣx²−(Σx)²][nΣy²−(Σy)²]} |
| Pearson covariance form | r = Cov(X,Y)/(σXσY), consistent conventions |
| Spearman (no ties) | ρ = 1 − 6Σd²/[n(n²−1)] |
| Spearman (with ties) | Pearson correlation of average ranks |
19. Summary and source scope
Correlation describes association, not causation. Pearson's r measures linear association and Spearman's rho measures association in ranks. Both require careful interpretation, particularly with outliers, nonlinear patterns, tied ranks and small samples. These are original supplementary study notes and constructed practice questions, not a verified reproduction of all official NCERT exercises. For exact textbook alignment consult https://ncert.nic.in/textbook.php and the applicable 2026, 27 CBSE curriculum at https://cbseacademic.nic.in/curriculum_2027.html.
Frequently asked questions
What is correlation?
Correlation summarises the direction and strength of association between two variables.
What is the difference between positive and negative correlation?
Positive correlation means variables tend to move together; negative correlation means one tends to increase as the other decreases.
What is Pearson's correlation coefficient?
A unit-free measure of linear association between two numerical variables, ranging from −1 to +1 when defined.
What is Spearman's rank correlation?
A measure of monotonic association based on ranks, suitable for ordinal or ranked numerical data.
How do you handle tied ranks in Spearman correlation?
Assign tied observations their average ranks, then calculate Pearson correlation on the resulting ranks.
Does correlation imply causation?
No. Confounding, reverse causation, chance and other explanations may produce correlation.
Can Pearson correlation be zero even when variables are related?
Yes. For example, a symmetric curved relationship can have zero linear correlation.
Are these official NCERT exercise solutions?
No. These are original supplementary notes and practice questions; consult the prescribed NCERT edition for exact exercises.
