Class 11 Humanities . Economics . Statistics For Economics . Correlation

Chapter 6: Correlation

Ch 6

Economics

Class 11 Humanities

Lecture 1

Introduction

Reference video · Lecture 1Introduction

Introduction to Correlation

Definition

Correlation analysis — A method for examining systematically whether and how two variables are related.

Examples of relationships

  • Temperature and visitors: Rising summer temperatures are associated with more visitors at hill stations.
  • Temperature and ice-cream sales: Ice-cream sales tend to rise as temperatures increase.
  • Tomato supply and price: When supply rises, prices may fall—for example, from Rs. 40 per kg to Rs. 4 per kg or less.
Seesaws illustrate whether two variables move together, move in opposite directions, and how strong their relationship is.
Seesaws illustrate whether two variables move together, move in opposite directions, and how strong their relationship is.

Types of Relationships

How relationships may arise
TypeMeaningExample
Cause-and-effect relationshipOne variable may influence or help explain changes in another.Low rainfall may be associated with low agricultural productivity.
Coincidental relationshipTwo variables change together, but there is no meaningful cause-and-effect explanation.Migratory birds arriving at a sanctuary and local birth rates; shoe size and money in a person's pocket.
Relationship due to a third variableA third factor affects both variables, making them appear related.Temperature increases both ice-cream sales and swimming-pool use, which may increase drowning deaths.

A third variable can create an apparent relationship

  1. Higher temperature → more ice-cream sales.
  2. Higher temperature → more people swim to cool off → more opportunities for drowning deaths.
  3. Ice-cream sales and drowning deaths may therefore be correlated, even though eating ice cream does not cause drowning.

What Correlation Measures

Definition

Correlation — A measure of the direction and intensity (strength) of the relationship between variables.

Core ideas

  • Covariation: Correlation describes how variables change in relation to one another.
  • Not causation: A correlation does not, by itself, prove that one variable causes a change in another.
  • Direction: When one variable changes, the other tends to change in the same direction or the opposite direction.
  • Linear assumption: In this chapter, an existing relationship is treated as linear—its relative movement can be represented by a straight line on a graph.
⚠ Common mistake

Do not interpret correlation as proof of cause and effect; coincidence or the influence of a third variable may explain the relationship.

Positive and Negative Correlation

Positive vs Negative Correlation
Positive correlationNegative correlation
Direction of movementVariables move in the same direction.Variables move in opposite directions.
When one variable risesThe other tends to rise too.The other tends to fall.
When one variable fallsThe other tends to fall too.The other tends to rise.
ExamplesIncome and consumption; temperature and ice-cream sales.Apple price and quantity demanded; study time and chances of failing.
Graphs illustrate positive and negative relationships between variables.
Graphs illustrate positive and negative relationships between variables.
Key terms
Correlation
The direction and strength of a relationship between variables.
Covariation
The way two variables change in relation to one another.
Positive correlation
A relationship in which variables move in the same direction.
Negative correlation
A relationship in which variables move in opposite directions.

What does correlation measure, and what does it not prove?

Answer: It measures the direction and intensity of covariation between variables; it does not prove a cause-and-effect relationship.

Distinguish positive from negative correlation.

Answer: In positive correlation, variables move in the same direction; in negative correlation, they move in opposite directions.

Quick recap

  • Correlation analysis examines how two variables are related.
  • A relationship may reflect cause and effect, coincidence, or the influence of a third variable.
  • Correlation measures the direction and intensity of covariation, not causation.
  • Positive correlation means variables move in the same direction.
  • Negative correlation means variables move in opposite directions.
  • The relationships considered here are assumed to be linear.
Lecture 2

Techniques For Measuring Correlation

Reference video · Lecture 2Techniques For Measuring Correlation

Main Techniques

Three ways to study correlation
TechniqueWhat it shows
Scatter diagramVisual direction and closeness of association; gives no numerical value.
Karl Pearson’s coefficientNumerical measure of the direction and degree of a linear relationship between two variables.
Spearman’s rank correlationMeasures linear association between ranks assigned to items according to their attributes.
Definition

Linear relationship — A relationship that can be represented by a straight line.

Definition

Attributes — Qualities that cannot be measured numerically, such as intelligence, physical appearance and honesty.

Scatter Diagram

Definition

Scatter diagram — A graph of paired values of two variables, plotted as points to assess their relationship visually without calculating a numerical measure.

Reading the pattern

  • Direction is shown by the overall trend; closeness of points indicates the strength of the relationship.
  • Positive correlation: points trend upward; as XX rises, YY tends to rise.
  • Negative correlation: points trend downward; as XX rises, YY tends to fall.
  • No correlation: points show no clear upward or downward trend.
  • Perfect correlation: all points lie exactly on a straight line; it may be perfectly positive or perfectly negative.
  • Linear pattern: points lie on or close to a straight line; a curved pattern is non-linear.
Scatterplot patterns: positive, negative, zero, perfect and non-linear correlation.
Scatterplot patterns: positive, negative, zero, perfect and non-linear correlation.

Karl Pearson’s Coefficient of Correlation

Definition

Karl Pearson’s coefficient — A precise numerical measure of the direction and degree of the linear relationship between two variables; also called the product-moment or simple correlation coefficient.

Means, deviations and covariance

  • Means: For NN paired observations, Xˉ=∑XN\bar X=\frac{\sum X}{N} and Yˉ=∑YN\bar Y=\frac{\sum Y}{N}.
  • Deviations: Let x=X−Xˉx=X-\bar X and y=Y−Yˉy=Y-\bar Y; these are deviations from the respective means.
  • Variance: σX2=∑(X−Xˉ)2N=∑X2N−Xˉ2\sigma_X^2=\frac{\sum (X-\bar X)^2}{N}=\frac{\sum X^2}{N}-\bar X^2; similarly for YY.
  • Standard deviation: σX\sigma_X and σY\sigma_Y are the positive square roots of the respective variances.
  • Covariance: Cov⁡(X,Y)=∑(X−Xˉ)(Y−Yˉ)N=∑xyN\operatorname{Cov}(X,Y)=\frac{\sum (X-\bar X)(Y-\bar Y)}{N}=\frac{\sum xy}{N}; its sign determines the sign of rr.
Pearson’s coefficient: deviation form
r=∑xyNσXσY=∑(X−Xˉ)(Y−Yˉ)∑(X−Xˉ)2∑(Y−Yˉ)2r=\frac{\sum xy}{N\sigma_X\sigma_Y}=\frac{\sum (X-\bar X)(Y-\bar Y)}{\sqrt{\sum (X-\bar X)^2}\sqrt{\sum (Y-\bar Y)^2}}

Use for paired observations; x=X−Xˉx=X-\bar X and y=Y−Yˉy=Y-\bar Y.

Pearson’s coefficient: shortcut form
r=N∑XY−(∑X)(∑Y)N∑X2−(∑X)2N∑Y2−(∑Y)2r=\frac{N\sum XY-(\sum X)(\sum Y)}{\sqrt{N\sum X^2-(\sum X)^2}\sqrt{N\sum Y^2-(\sum Y)^2}}

Use when the sums of raw values, squares and cross-products are available.

⚠ Common mistake

Use Pearson’s coefficient only after checking that the relationship is linear. For a non-linear pattern, the coefficient can be misleading.

Properties and Interpretation of rr

  • Range: −1≤r≤1-1\leq r\leq 1. A calculated value outside this range signals an error.
  • No units: rr is a pure number; its value does not include the units of measurement.
  • Sign: r>0r>0 means the variables move in the same direction; r<0r<0 means they move in opposite directions.
  • Strength: Values near +1+1 or −1-1 show a strong linear relationship; values near 00 show a weak linear relationship.
  • Zero: r=0r=0 means no linear relationship, although a non-linear relationship may exist.
  • Perfect relationship: r=1r=1 or r=−1r=-1 indicates an exact linear relationship.
  • Origin and scale: Changing origin and scale does not change rr when the scale factors are of the same sign.
Change of origin and scale
U=X−Ah,V=Y−Ck,rXY=rUVU=\frac{X-A}{h},\quad V=\frac{Y-C}{k},\quad r_{XY}=r_{UV}

If A,CA,C are assumed means and h,kh,k are common scale factors of the same sign, the transformed variables have the same correlation as the originals.

Examples and interpretation

  • Positive examples: A rise in coffee prices may increase demand for tea; better irrigation may raise yield; higher temperatures may increase ice-cream sales.
  • Negative examples: A rise in a commodity’s price may lower its demand; higher interest rates may reduce demand for funds.
  • Context matters: Deaths and doctors sent to villages may be positively correlated because doctors are sent to the worst-affected places; some deaths may be terminal or unrelated to the epidemic.
  • Delayed effects: Doctors’ benefits may appear later, and events such as a tsunami can increase deaths independently of the epidemic.
  • English and Statistics marks: r=0.1r=0.1 shows weak positive association, while r=0.9r=0.9 shows strong positive association.
  • Vegetable supply and price: r=−0.9r=-0.9 indicates a stronger inverse relationship than r=−0.1r=-0.1; a large supply may not lower price if transport sends it to other markets.
Farmer education and annual yield
Years of schooling (XX)Yield per acre in '000 Rs. (YY)
04
24
46
610
810
108
127

Example calculation: farmer education and yield

  • Means: For the seven pairs, Xˉ=6\bar X=6 and Yˉ=7\bar Y=7.
  • Deviation totals: ∑(X−Xˉ)2=112\sum (X-\bar X)^2=112, ∑(Y−Yˉ)2=38\sum (Y-\bar Y)^2=38 and ∑(X−Xˉ)(Y−Yˉ)=42\sum (X-\bar X)(Y-\bar Y)=42.
  • Result: r=4211238=0.644r=\frac{42}{\sqrt{112}\sqrt{38}}=0.644; education and yield are positively correlated, with a fairly strong linear association.

Step Deviation Method

Definition

Step deviation method — A shortcut for calculating correlation by transforming large values with assumed means and common factors; the correlation remains unchanged.

Calculate correlation using transformed values

  1. Choose assumed means A,CA,C and convenient scale factors h,kh,k; transform each paired observation.
    U=X−Ah,V=Y−CkU=\frac{X-A}{h},\qquad V=\frac{Y-C}{k}
  2. Calculate ∑U\sum U, ∑V\sum V, ∑U2\sum U^2, ∑V2\sum V^2, ∑UV\sum UV and the number of pairs NN.
  3. Substitute the transformed sums in the shortcut formula; this gives the original correlation.
    r=N∑UV−(∑U)(∑V)N∑U2−(∑U)2N∑V2−(∑V)2r=\frac{N\sum UV-(\sum U)(\sum V)}{\sqrt{N\sum U^2-(\sum U)^2}\sqrt{N\sum V^2-(\sum V)^2}}
Step-deviation example: price index and money supply
U=(X−100)/10U=(X-100)/10V=(Y−1700)/100V=(Y-1700)/100U2U^2V2V^2UVUV
21412
5325915
98816472
1210144100120
1313169169169
Totals: ∑U=41\sum U=41∑V=35\sum V=35∑U2=423\sum U^2=423∑V2=343\sum V^2=343∑UV=378\sum UV=378
Step-deviation example result
r=378−41×355423−4125343−3525=0.98r=\frac{378-\frac{41\times35}{5}}{\sqrt{423-\frac{41^2}{5}}\sqrt{343-\frac{35^2}{5}}}=0.98

For the five price-index and money-supply pairs, the transformed sums give a strong positive relationship.

  • Interpretation: Price index and money supply have a strong positive correlation; an increase in money supply is associated with a rise in the price index.

When should Karl Pearson’s coefficient be used?

Answer: Use it when the relationship between the variables is linear. Check the scatter diagram first; a non-linear pattern can make the result misleading.

What does r=0r=0 mean?

Answer: There is no linear relationship between the variables; a non-linear relationship may still exist.

Quick recap

  • Scatter diagrams show direction and closeness visually but give no numerical coefficient.
  • Karl Pearson’s coefficient measures the direction and strength of a linear relationship.
  • The sign of rr shows direction; its magnitude shows strength, and −1≤r≤1-1\leq r\leq 1.
  • A zero Pearson coefficient means no linear association, not necessarily no relationship of any kind.
  • Check the scatter diagram and interpret correlations in context before drawing conclusions.
  • The step deviation method simplifies calculations without changing the correlation.
Lecture 3

Spearman’s rank correlation

Reference video · Lecture 3Spearman’s rank correlation

Meaning and uses

Definition

Spearman’s rank correlation — A measure of linear association between the ranks assigned to observations, rather than between their original values.

When it is useful

  • Origin: Developed by British psychologist C.E. Spearman.
  • Unmeasurable values: Use ranks when measurements are unavailable, such as ranking students by height and weight without instruments.
  • Relative qualities: Use it for qualities such as fairness, honesty or beauty when at least one variable can only be judged relatively.
  • Clear-direction non-linear relationships: It can be used in some cases where the direction of association is clear but the relationship is non-linear.
  • Extreme values: It is not affected by extreme values, making it useful in such data and, in this respect, preferable to Karl Pearson’s coefficient.

Formula and interpretation

  • Calculation idea: Replace each observation with its rank, then compare the paired ranks for the two variables.
  • Rank difference (DD): The difference between the ranks assigned to the same observation for the two variables.
  • Interpretation: Like the simple correlation coefficient, it indicates the direction and degree of association; here, the association is between ranks.
  • Range: The coefficient lies between −1-1 and +1+1.
  • Limitation: It is generally less accurate than the ordinary method because ranking does not use all the information in the original data.
Spearman’s rank correlation coefficient
rk=1−6∑D2n3−nr_k = 1 - \frac{6\sum D^2}{n^3-n}

Use for nn observations; DD is the difference between the paired ranks.

Relation to the simple correlation coefficient

  • First differences: Differences between consecutive values after arranging the series in order of magnitude.
  • Usual pattern: First differences are almost never constant; values often cluster around the centre, where differences tend to be smaller.
  • If first differences were constant: The simple correlation coefficient rr and rank coefficient rkr_k would give identical results.
  • General relation: In general, rkr_k is less than or equal to rr.

Quick recap

  • Spearman’s rank correlation measures linear association between ranks, not the original values.
  • It is useful for unmeasurable or relative qualities, some clear-direction non-linear relationships, and data with extreme values.
  • Its formula is rk=1−6∑D2n3−nr_k = 1 - \frac{6\sum D^2}{n^3-n}, where DD is the paired rank difference and nn is the number of observations.
  • It ranges from −1-1 to +1+1 and has the same general interpretation as the simple correlation coefficient.
  • It usually uses less information and is less accurate than the ordinary method; in general, rk≤rr_k \leq r.
Lecture 4

Calculation of Rank Correlation Coefficient

Reference video · Lecture 4Calculation of Rank Correlation Coefficient
Definition

Spearman’s rank correlation coefficient — A measure of the direction and strength of the linear relationship between two variables using their ranks.

Case 1: Ranks are given

Spearman’s rank correlation (no ties)
rs=1−6∑D2n(n2−1)r_s=1-\frac{6\sum D^2}{n(n^2-1)}

Here, DD is the difference between paired ranks and nn is the number of pairs.

Judges’ rankings
JudgeCompetitor 1Competitor 2Competitor 3Competitor 4Competitor 5
A12345
B24153
C13524

Comparing judges’ rankings

  • Compare pairs: Three judges produce three possible pairs; calculate rank correlation for each pair.
  • A and B: ∑D2=14\sum D^2=14, n=5n=5; therefore rs=1−6(14)5(52−1)=0.30r_s=1-\frac{6(14)}{5(5^2-1)}=0.30.
  • A and C: ∑D2=10\sum D^2=10, n=5n=5; therefore rs=0.50r_s=0.50.
  • B and C: The listed ranks give ∑D2=28\sum D^2=28, so rs=1−6(28)120=−0.40r_s=1-\frac{6(28)}{120}=-0.40.
  • Conclusion: A and C have the closest rankings because their coefficient, 0.500.50, is the largest.
⚠ Common mistake

The source states that the B–C coefficient is 0.90.9, but the listed rankings produce −0.4-0.4. Applying the formula to the given ranks shows that the stated value is inconsistent with the data.

Case 2: Ranks are not given

Rank the data, then calculate

  1. Rank each variable separately, using the same convention for both; in this example the highest mark receives rank 1.
  2. For each student, find the difference between the two ranks and square it.
    D=R1−R2D=R_1-R_2
  3. Substitute ∑D2\sum D^2 and nn into Spearman’s formula.
Students’ marks and ranks
StudentStatistics marksEconomics marksStatistics rankEconomics rank
A856011
B604845
C554954
D655033
E755522

Students’ example result

  • Rank differences: Only students B and C have non-zero differences; each has D2=1D^2=1, so ∑D2=2\sum D^2=2.
  • Coefficient: For n=5n=5, rs=1−6(2)5(52−1)=0.90r_s=1-\frac{6(2)}{5(5^2-1)}=0.90.

Case 3: Ranks are repeated

Assigning tied ranks

  • Common rank: Give tied values the average of the rank positions they would have occupied.
  • Next rank: After a tie, continue with the next rank after all positions used by the tied values.
  • Example: Three equal values spanning ranks 9, 10 and 11 each receive rank 1010.
Rank correlation with ties
rs=1−6[∑D2+∑jmj3−mj12]n(n2−1)r_s=1-\frac{6\left[\sum D^2+\sum_j\frac{m_j^3-m_j}{12}\right]}{n(n^2-1)}

For each tied group, mjm_j is the number of repeated values. Add every group’s correction factor to ∑D2\sum D^2.

Example with repeated values

  • Tied groups in YY: The value 5050 occurs three times, while 9090 and 7575 each occur twice.
  • Correction factors: For these groups, the factors are 33−312=2\frac{3^3-3}{12}=2, 23−212=0.5\frac{2^3-2}{12}=0.5, and 23−212=0.5\frac{2^3-2}{12}=0.5; total correction is 33.
  • Rank differences: The rank table gives ∑D2=198\sum D^2=198 and n=12n=12.
  • Coefficient: rs=1−6(198+3)12(122−1)≈0.30r_s=1-\frac{6(198+3)}{12(12^2-1)}\approx0.30.
  • Interpretation: XX and YY have a positive but not strong rank relationship.

Choosing and interpreting correlation

Pearson’s correlation vs rank correlation
PointKarl Pearson’s coefficientSpearman’s rank coefficient
BasisMeasured variable valuesRanks of the values
UseNumerically measured variablesVariables that cannot be measured precisely or are expressed as ranks
Relationship measuredLinear relationshipLinear relationship using ranks

What correlation tells us

  • Scatter diagram: Gives a visual presentation of the nature of a relationship and is not limited to linear relationships.
  • Correlation coefficients: Karl Pearson’s coefficient and Spearman’s rank coefficient measure linear relationships numerically.
  • Interpretation: Correlation indicates the direction and intensity of change associated with another variable; it does not establish causation.
★ Remember
  • Rank method: Rank both variables consistently; when ranks tie, use average ranks and include all required correction factors.
  • Correlation is not causation: It shows that variables vary together, not that one causes the other.

Quick recap

  • Spearman’s coefficient measures direction and strength of a relationship using ranks.
  • With no ties, calculate rank differences, square them and apply the rank-correlation formula.
  • When ranks are missing, rank each variable first using a consistent order.
  • Tied values receive the average of the rank positions they occupy.
  • Add a correction factor for every tied group when ranks are repeated.
  • Scatter diagrams show relationships visually; correlation does not establish causation.

Economics — ALL CHAPTERS

1

Introduction

View

2

Theory of Consumer Behaviour

View

3

Production And Costs

View

4

The Theory Of The Firm Under Perfect Competition

View

5

Market Equilibrium

View

1

Introduction

View

2

Collection of Data

View

3

Organizing of Data

View

4

Presentation of Data

View

5

Measures of Central Tendency

View

6

Correlation

View

7

Index Numbers

View

8

Use of Statistical Tools

View