Do Mega Rallies by Political Leaders Matter?
Theory and Evidence from India’s General Elections
2025-07-30
Research Question
- Political rallies remain a key campaign tool, especially rallies by top leadership.
- We ask two questions:
- Do rallies mobilize electoral support (vote share or turnout)?
- Are they strategically deployed?
- Findings:
- Rallies moderately affect turnout but effect on support is marginal and highly contingent.
- They have strong indirect effects through increasing awareness about the party platform.
- Their employment is only semi-strategic, not conforming to conventional expectations, and highly contingent on the party.
Theoretical Framework
- Rallies blend the mass reach of social media with personalized appeal of face-to-face campaigns.
- Particularly relevant in leader-centric, weakly-institutionalized party systems like India
- Key Mechanisms through which Rallies can have an effect:
- Direct effects: boost turnout or vote share.
- Indirect effects: raise awareness of party messages, nationalize campaign.
- But these rallies are expensive events. Their deployment must factor in the rational calculus of expected gains and costs of placing it in one constituency and not the other.
Data
- Data Collected
- Rally data from party websites, The Hindu, and YouTube.
- Lok Sabha results: 2009–2024 (LokDhaba & ECI).
- Voter awareness data: NES Post-Poll 2019.
- Candidate characteristics from ADR.
- Exploratory Data/ Collecting
- Assembly segment Level Rally Locations
- Qualitative interviews with party functionaries.
Election Schedule 2019 and 2024
Rallies in 2019 and 2024
Distribution of Modi and Rahul Gandhi’s Rallies Across Parliamentary Constituencies in 2019 and 2024
Frequency of Rallies in 2019 by Phases
Frequency of Rallies in 2024 by Phases
Rallies in 2019 and 2024
Do rallies increase turnout?
- Hypothesis: The conduct of a rally by a party’s top leadership in a given constituency increases voter turnout within that constituency.
Two-Way FE Estimates of Rally Effects on Turnout Change
Do rallies increase support?
- Hypothesis: The conduct of a rally by a party’s top leadership in a given constituency increases voter turnout within that constituency.
OLS Estimates of Rally Effects on Party Vote Swing
Do rallies have indirect effects?
- Hypothesis: The conduct of a rally by a party’s top leadership in a given constituency increases public familiarity with the party’s rhetoric.
Share Voting Based on PM Candidate by Any Rally Presence (NES 2019)
Do rallies have indirect effects?
Effect of Rally Presence on Awareness of Party’s Salient Issue (NES 2019)
Deploying rallies
Hypothesis 1: Mega Rallies by top leadership are less likely to be placed in party strongholds.
Hypothesis 2: Mega Rallies by top leadership are more likely to be placed in closely contested constituencies.
Deploying rallies
Logistic Regression Estimates of Rally Placement by Party Strength and Candidate Characteristics
Conclusion
- Key findings:
- In line with earlier research, rallies have limited direct electoral impact in terms of turnout and vote share.
- Rallies possess indirect discursive power, helping to shape and reinforce the campaign agenda by creating greater awareness and nationalizing the competetion.
- On deployment:
- Limited evidence for strategic placement of rallies based on electoral considerations.
- Stronger evidence for organizational and structural determinants, such as party infrastructure and local conditions.
- Conclusion:
- Rallies act more as public spectacles or rituals essential to campaigning.
- Their placement reflects structural, exogenous constraints rather than purely strategic choices by parties.
Future Plans
Collecting Micro level Data: These findings are contingent on the data currently available, and we believe that a more fine-grained analysis—using smaller units such as assembly segments within parliamentary constituencies, may reveal a clearer and more accurate direct effect of political rallies. We are in the process of identifying exact rally locations to facilitate this.
Spatial Econometrics: We aim to further estimate the spillover effect of the rallies in neighboring segments by incorporating methods such as spatial regression.
Qualitative insights from interviews with relevant individuals.
Early Results
Effect of Rallies on Turnout
Appendix
Turnout Specification
Test:
\[ \Delta Turnout_{ct} = \beta_1 Rally_{ct} + \beta_2 Turnout Change_{ct} + \alpha_c + \lambda_t + \epsilon_{ct} \]
\(\Delta Turnout_{ct}\): Change in voter turnout in constituency \(c\) at time \(t\) compared to last election.
\(Rally_{ct}\): A binary variable = 1 if a top leader held a rally in constituency \(c\) during the campaign period.
\(TurnoutChange_{ct}\): Change in turnout between the two prior elections in constituency \(c\), used to account for pre-existing turnout trends.
\(\alpha_c\) (Constituency fixed effects): Controls for time-invariant characteristics of constituency \(c\) (e.g., demographics, geography).
\(\lambda_t\) (Year fixed effects): Controls for election-year-specific factors that affect turnout across all constituencies.
Turnout Discussion
- Constituencies with any rally saw an average increase of 1.35 percentage points in turnout (p < 0.001).
- When separated by leader:
- Rahul Gandhi’s rallies led to a 1.66 pp increase in turnout (p < 0.01).
- Narendra Modi’s rallies led to a 1.03 pp increase (p < 0.05).
- In the fully specified model (with both leaders included):
- Effects of Rahul’s and Modi’s rallies remain positive but lose statistical significance.
- Rahul’s effect is still marginally significant (p < 0.1), aligning with the hypothesis.
- The ‘any leader’ rally effect is near zero and statistically insignificant.
- Across models, prior turnout change shows a strong negative and significant coefficient, indicating mean reversion in participation.
- Overall, results provide suggestive evidence that Rahul Gandhi’s rallies are more consistently linked to increased turnout.
Vote Share Model Specification
Test:
\[ \Delta VoteShare^{Party}_{ct} = \beta_1 Rally^{Leader}_{ct} + \beta_2 Stronghold^{Party}_{ct} + \mathbf{X}_{ct}'\boldsymbol{\gamma} + \epsilon_{ct} \]
\(\Delta VoteShare^{Party}_{ct}\): Change in the vote share of the party in constituency \(c\) at time \(t\) compared to the previous election.
\(Rally^{Leader}_{ct}\): A binary indicator equal to 1 if the party’s top leader held a rally in constituency \(c\) during the campaign period.
\(Stronghold^{Party}_{ct}\): A binary indicator equal to 1 if the constituency is a historical stronghold of the party.
\(\mathbf{X}_{ct}'\): A vector of control variables (e.g., turnout change, demographic controls).
\(\boldsymbol{\gamma}\): Coefficients for the control variables.
Vote Share Model Discussion
- Limited evidence that rallies by national leaders systematically increase party vote share.
- For INC in 2024, Rahul Gandhi’s rallies are associated with a +3.69 percentage point increase (p < 0.05).
- For INC in 2019, the effect is positive but not significant (+2.09 pp, p > 0.1).
- For BJP, Modi’s rallies show:
- No significant effect in 2024 (–0.42 pp, p > 0.1),
- A modest, positive effect in 2019 (+0.97 pp), also not statistically significant.
- In 2024, opponent stronghold is associated with a +5.18 pp increase in INC support (p < 0.001), suggesting strategic targeting.
- Conclusion: Rallies may amplify support in favorable contexts (e.g., INC in 2024), but do not consistently generate vote share gains on their own.
Issue Awarenes Model Specification
Test:
\[ \text{Issue Awareness}_{ic} = \beta_1 \, \text{Rally}_{c} + \mathbf{X}_{ct}'\boldsymbol{\gamma} + \epsilon_{ct} \]
\(\text{Issue Awareness}_{ic}\): A binary indicator equal to 1 if the voter was familiar with the flagship concern/narrative of the party.
\(\text{Rally}_{c}\): Indicates whether a high-profile rally by the party leader occurred in constituency (c).
\(\mathbf{X}_{ct}'\): A vector of individual covariates (e.g., interest in elections, rally attendance, and whether a party member visited their house).
\(\boldsymbol{\gamma}\): Coefficients for the control variables.
Issue Awareness Model Discussion
- The results indicate:
- Attending a rally and having one in the respondent’s constituency both significantly increase awareness of the party’s key campaign issue.
- These effects remain robust even after controlling for:
- Partisanship
- Party worker visits
- Political interest concerns.
- Partisanship
- This visibility elevates these issues as the defining concerns of the campaign.
- Key implication:
- Rallies not only highlight key issues but also generate demand for them by:
- Broadening public exposure.
- Reducing reliance on sustained grassroots mobilization.
- Replacing it with sporadic, high-visibility events that:
- Frame key issues
- Create interest and demand for them
- Rallies not only highlight key issues but also generate demand for them by:
Rally Deployment Specification
Deploying Rallies
Test:
\[ \text{Pr}(\text{Rally}_{it} = 1) = \text{logit}^{-1}\Big( \beta_0 + \beta_1 \cdot \text{Stronghold}_{it} + \\ \beta_2 \cdot \text{CloseComp}_{it} + \beta_3 \cdot \text{Reservation}_{it} + \\ \beta_4 \cdot \text{StateGovt}_{it} + \beta_5 \cdot \text{Assets}_{it} + \\ \beta_6 \cdot \text{CriminalCases}_{it} + \varepsilon_{it} \Big) \]
\(\text{CloseComp}_{it}\): A binary indicator equal to 1 if the constiteuncy was a close competetion for the party in the last election
\(\text{Reservation}_{it}\): Indicates whether it was a reserved constituency.
\(\text{StateGovt}_{it}\): Indicates whether the party exercises majority in the state assembly.
\(\text{Assets}_{it}\): Log of total assets of the local candidate
\(\text{CriminalCases}_{it}\) : Log of total criminal cases against the local candidate
Rally Deployment Discussion
- As opposed to conventional wisdom:
- Rallies are not consistently deployed on stronghold status
- Margin of victory/loss seems to have no impact on the placement of rallies
- Reservation of seats does not affect rally placement.
- Leader Based Differences:
- Rahul Gandhi was much more likely to conduct a rally in a state where INC was in power in 2019.
- Rahul Gandhi appears to have preferred conducting rallies in constituencies with more resourceful local candidates
- Rahul Gandhi appears to have conducted more rallies in constituencies with local candidates having more criminal cases against them.
- Mean number of criminal cases against alliance candidates was notably higher in 2024 than in 2019.
- Rally placement reflects party strength and organizational constraints more than universal vote-maximization logic as was hypothesized before.
Delhi School on Political Analytics, 2024