-
Voting Intention in Colombia
Ahead of the presidential elections in Colombia, the survey conducted by GAD3 for RCN Televisión indicates that 34% of respondents would vote for Iván Cepeda, representative of the Pacto Histórico. He is followed by Abelardo de la Espriella, with 26% support, and Paloma Valencia, candidate of the Centro Democrático, with 4%
-
In Potential Run-off Scenarios
We asked respondents about their voting intention across different presidential run-off scenarios.
In a potential match-up between Cepeda and Abelardo de la Espriella, Cepeda would obtain 38% of the vote, compared to 33% for De la Espriella.
If the run-off were between Cepeda and Fajardo, the results would be 36% and 26%, respectively.
In a hypothetical scenario between Cepeda and Paloma Valencia, Cepeda would reach 40%, while Paloma Valencia would obtain 25%.
On the other hand, if the contest were between Abelardo de la Espriella and Fajardo, De la Espriella would secure 31% compared to 20% for Fajardo.
Finally, in a potential run-off between Paloma Valencia and Abelardo de la Espriella, the figures would be 13% and 34%, respectively.
Narciso Michavila, President of GAD3, took part in a panel on Noticias RCN to analyse the key findings of the survey. Here is a clip from his participation
Technical Details:
Universe: Residents in Colombia aged 18 and over with the right to vote in the 2026 electoral processes and with access to mobile telephony.
Geographical scope: Nationwide coverage across the continental territory (31 departments and the Capital District of Bogotá). The sampling design ensures territorial representativeness and the inclusion of the most populous municipalities, in compliance with Article 4 of Law 2494 of 2025.
Data collection procedure: Telephone surveys (CATI) conducted by trained interviewers using software-assisted systems.
Sample size: 2,473 nominal interviews. Effective sample size after weighting (Kish): 1,772 interviews.
Margin of error: ±2.3% for the total sample (n = 2,473) at a 95% confidence level and P = Q = 0.5, based on the calculated effective sample size. The effective number of responses (n-effective) is calculated using the Kish formula to ensure the accuracy of the published confidence intervals, guaranteeing that for all voting indicators (Art. 4, Law 2494) the margin of error does not exceed 3.0%. Specific error calculations, based on the effective sample size after weighting, are detailed in the Extended Methodological Report provided to ensure full verifiability of the study.
Weighting of the sample: Representativeness adjustment through post-stratification calibration (raking) in two stages. First stage: national-level adjustment for sociodemographic variables (sex, age and socio-economic stratum) according to DANE 2025 projections. Second stage: macro-regional adjustment for electoral behaviour (2022 vote recall) based on official data from the National Registry. The process includes reinforcement of political likelihood ratios and variance control through the limitation of extreme weights to maximise sample efficiency (effective n).
Confidence level: 95%, under the most unfavourable assumption of p = q = 0.5 and simple random sampling.
Sampling frame: Territorial selection is carried out proportionally to population size according to the National Population and Housing Census (DANE). In compliance with Law 2494 of 2025, the sampling design ensures the inclusion of the largest municipalities and districts, as well as coverage of key regional municipalities where no municipality or district reaches a population of 800,000 inhabitants. This criterion is applied solely to ensure adequate territorial representativeness of the study.
Interview duration: Approximately 4–5 minutes.
Fieldwork dates: 16 to 23 February 2026.
Legal entity responsible for the study: GAD3 Colombia S.A.S.
Professionals conducting the survey: Dr Narciso Michavila Núñez, Javier Sánchez García and Mario Martín Fernández.
