Standard methodology
Overview
SCORE conducts pre-election opinion polls among registered voters aged 18 and above across Assembly constituencies. These polls assess voting intentions, assessments of public representatives and governments, and opinions on issues of public importance. Poll-specific coverage, fieldwork dates, interview counts and languages are reported with each release.
Sampling and data collection
SCORE uses a multi-stage, stratified quota sampling approach, covering constituencies and distributing interviews across rural and urban localities. Respondent quotas reflect relevant demographic characteristics. Data are collected through live telephone and face-to-face interviews conducted by trained interviewers. This is a non-probability sample: respondents are not randomly selected and their probabilities of selection are unknown.
Weighting and analysis
Poll responses are weighted against available demographic and electoral benchmarks, including past election results where appropriate. Telephone and face-to-face samples are assessed and combined analytically. Vote-intention estimates may incorporate statistical modelling of undecided voters and reported voting patterns. Where published forecasts include analyst adjustments, these are distinguished from poll-based estimates.
Seat projections
Where provided, constituency-level projections draw on local poll responses, demographic modelling, previous election results and relevant constituency information. Simulation and field review may inform final seat assessments. Seat projections are forecasts, not direct poll measurements or guaranteed outcomes.
Quality assurance
Responses are reviewed for completeness, duplication, geographic consistency and coding accuracy. Face-to-face interviews are checked using available location and time records; records that fail applicable validation checks may be excluded. The precise checks and exclusions for each poll are documented in its technical records.
Interpretation and limitations
Findings reflect stated opinions during the fieldwork period and may change before an election. Quota sampling, non-response, coverage gaps, response bias, weighting assumptions and modelling choices can affect estimates. Conventional probability-sampling margins of error do not apply directly to this non-probability design; any uncertainty ranges require poll-specific explanation. Small constituency samples and close contests warrant particular caution.
Disclosure and data protection
Each release should identify its target population, geographic coverage, fieldwork dates, sample size, interview modes and relevant methodological departures. Personal identifiers used for verification are kept separate from published responses; individual respondents are not identified in reporting. Publication is subject to applicable electoral and data-protection requirements.