Election Dataset Documentation (India, 1991–2023)

1. Overview

This dataset compiles comprehensive electoral data for both state-level legislative assembly elections and national-level general (Lok Sabha) elections in India. It covers all 29 Indian states(Present 28 states and a former state Jammu & Kashmir) and parliamentary constituencies from 1991 to 2023. The dataset is structured as a balanced panel, with one row per State × Year (or Nation × Year, in the case of parliamentary elections), enabling longitudinal analysis of political trends, incumbency dynamics, and electoral outcomes across time and space.

All data have been meticulously compiled from the official Election Statistical Reports published by the Election Commission of India (ECI).

2. Unit of Observation

• For state elections:
  One row = One Indian state × One calendar year
• For national elections:
  One row = India-wide outcome × One calendar year

Time period: 1991 to 2023

3. Variables

4. Special Cases and Coding Decisions

• Coalition governments:
  • Winning_Party/Alliance = Full name of the alliance (e.g., “INC+NCP”)
  • Ruling_Party = Largest party within that alliance (e.g., “INC”)
• President’s Rule:
  • Ruling_Party = “President Rule”
  • All other election-related variables for that year are left blank
• General (Parliamentary) Elections:
  • The same structure and variables are applied to national-level data
  • State is coded as “All India” for these entries

5. Data Source

All data are extracted from the Election Commission of India’s Election Statistical Reports, available in the public domain. These are official documents that include constituency-level results and aggregate summaries for every Indian state and parliamentary election.

6. Structure and Data Quality

• Dataset is balanced: 30( all 29 states + India) × 33 years (1991–2023) = 990 observations at the state level
• Consistent formatting and spelling for all party and alliance names
• All vote figures are in absolute numbers, not percentages
• Handles edge cases like mid-term elections, early dissolutions, and President’s Rule uniformly

7. Public Use and Research Applications

This dataset is intended for open-access academic and public research. It enables a wide range of empirical investigations into the political economy of Indian elections. Researchers may use it to study incumbency advantage, the effect of party competition on vote shares, or the evolution of regional vs. national parties over time. It can be readily merged with fiscal, demographic, or policy datasets to support econometric modeling, event-study analysis, and comparative political research. The dataset’s panel structure, organized by state and year, enables rigorous empirical analysis of temporal and cross-sectional variation. The staggered scheduling of elections across states introduces natural variation over time, allowing for the identification of dynamic patterns and potential causal relationships in electoral and policy-related outcomes.