Write short notes on the following in about 150 words each :
Correlation and Regression
10 marks · answer in 150 words · directive: Write short notes
The core of a good answer
Correlation measures the strength and direction of a linear association between two variables with a symmetric coefficient r (from minus 1 to plus 1), while regression gives the equation that predicts one variable from another (Y = a + bX) with a slope that is not symmetric; the two are linked by b = r × (sY/sX) and bYX × bXY = r².
An outline written by upsc.help, not an official UPSC answer; every fact checked by a separate review. How we work
What the question asks
A short note on two biostatistical tools: define each, give the formulas, contrast them, and show how they are used in botany and plant breeding. A small worked example or a scatter diagram with the fitted line makes the note concrete.
Answer outline
Introduction
Both deal with paired observations (X, Y) on the same units, for example nitrogen dose and plant height, or leaf area and yield.
Correlation
- Pearson's coefficient r = Σ(x − x̄)(y − ȳ) / √[Σ(x − x̄)² × Σ(y − ȳ)²].
- r lies between −1 and +1; r = 0 means no linear relation (not necessarily independence); r is unitless and symmetric (rXY = rYX).
- Types: positive or negative, simple, partial or multiple; Spearman's rank correlation for ranked or non-normal data.
- Correlation does not prove causation; both variables may depend on a third factor.
Regression
- Simple linear regression line of Y on X: Y = a + bX, fitted by least squares.
- b (regression coefficient) = Σ(x − x̄)(y − ȳ) / Σ(x − x̄)²; a = ȳ − b x̄.
- Two regression lines exist (Y on X and X on Y); they cross at (x̄, ȳ) and coincide only when r = ±1.
- r² (coefficient of determination) is the fraction of variation in Y explained by X.
Worked illustration
- N dose X (kg/ha): 0, 20, 40, 60, 80; plant height Y (cm): 50, 56, 61, 65, 68.
- x̄ = 40, ȳ = 60; Σ(x − x̄)(y − ȳ) = 900; Σ(x − x̄)² = 4000; Σ(y − ȳ)² = 206.
- r = 900 / √(4000 × 206) = 0.991; b = 900/4000 = 0.225 cm per kg/ha; a = 60 − 0.225 × 40 = 51.0.
- Y = 51.0 + 0.225X predicts 62.25 cm at 50 kg/ha; r² = 0.983.
Uses in plant science
- Correlation among yield components guides indirect selection; path coefficient analysis (Sewall Wright) splits correlations into direct and indirect effects.
- Parent and offspring regression estimates narrow sense heritability (h² = b on mid-parent value; 2b on one parent).
- Dose and response curves, growth analysis and calibration (standard curves in spectrophotometry) use regression.
Conclusion
Correlation tells how closely two characters move together; regression tells how much one changes per unit of the other and allows prediction.
Examples and data to cite
- Francis Galton introduced 'regression' in his 1886 paper 'Regression towards mediocrity in hereditary stature', on parent and child heights.
- Karl Pearson formalised the product moment correlation coefficient in the 1890s.
- Example computed above: r = 0.991 between nitrogen dose and plant height for the five paired values given.
Facts to remember
- b(YX) = r × sY/sX and b(XY) = r × sX/sY, so b(YX) × b(XY) = r² and r = ±√(b(YX) × b(XY)) with the sign of the coefficients.
- Regression coefficients have units; r has none.
- Correlation and regression coefficients are unaffected by change of origin; r is also unaffected by change of scale.
- Significance of r is tested with t = r√(n − 2)/√(1 − r²) on n − 2 degrees of freedom.