Correlation, Regression and Index Numbers: The Applied Statistics Toolkit
Commerce2 min readNov 21, 2026

Correlation, Regression and Index Numbers: The Applied Statistics Toolkit

Correlation, Regression and Index Numbers: The Applied Statistics Toolkit
2 min read · 341 words

Quick Answer

Correlation measures the strength and direction of association (Pearson’s r from −1 to +1); regression fits a predictive line ŷ = a + bx by least squares. r² is the explained-variance share. Index numbers, time-series components and extrapolation complete the applied toolkit.

Visual Summary

  • Pearson r — linear association from −1 to +1.
  • r² — proportion of variance the model explains.
  • Least squares — minimises squared vertical deviations.
  • Laspeyres — base-year quantity weights in index numbers.

Correlation and Regression

Pearson’s r measures linear association; Spearman’s rho ranks ordinal data. The regression slope and r share a sign; r² (coefficient of determination) states how much of y’s variance the model explains. Least squares minimises the sum of squared vertical deviations, making big errors costly. Significance of the slope tests association — never proof of causation; extrapolating beyond the observed range risks relationships that change outside the data.

Index Numbers

Index numbers track relative change with the base year at 100. Laspeyres weights price changes by base-year quantities (cost of the old basket today); Paasche uses current-year quantities; Fisher’s ideal index is their geometric mean. Consumer-price and wholesale-price indices apply these constructions to inflation measurement.

Time-Series Components

The classical decomposition: Trend (long-term direction), Seasonal (within-year calendar rhythm), Cyclical (multi-year waves) and Irregular (random noise) — combined additively (Y = T + S + C + I) or multiplicatively (Y = T × S × C × I) depending on whether variation scales with the level. Forecasting beyond observed data is extrapolation; smoothing (moving averages) separates trend from noise.

Exam Angle

NET passages test the r/r² distinction, the Laspeyres-Paasche weights, and the four components. Practise in the UGC NET Commerce mock.

FAQ

Does correlation prove causation?
Never — association describes co-movement; causation needs design or theory.

Which index uses current-year weights?
Paasche. Laspeyres uses base-year quantities.

Related reading

Exam angle

How exams use this: Prelims tests Economy via scheme–ministry–outlay pairs, Articles and macro numbers; one confused figure costs the question. Test it under exam conditions →

Quick revision

  • Pearson r: — linear association from −1 to +1.
  • r²: — proportion of variance the model explains.
  • Least squares: — minimises squared vertical deviations.
  • Laspeyres: — base-year quantity weights in index numbers.
  • UPSC Economy: Complete Topic-Wise Weightage Guide and 3-Month Study Path
  • National Income Explained: GDP, GNP, NNP, NDP at Factor Cost vs Market Price for UPSC Economy
ShareTelegramX

Have a doubt on this topic?