Estonia's Polling Institutes: Comparison and Predictive Accuracy

The PolitPro Score assesses the reliability of polling institutes by comparing their data with actual election results and the calculated trend. A high score reflects an accurate representation of political sentiment without systematic distortion. Significant, repeated deviations for individual parties lead to point deductions. Such discrepancies often indicate "house effects"—methodological peculiarities in weighting that systematically over- or under-represent certain parties. The score provides transparent guidance on which institutes most accurately mirror reality. The maximum score is 100.

Accuracy indicates how closely pre-election polls aligned with actual election results. We compare final surveys from various institutes with official outcomes. To do this, we calculate the average deviation across all parties that received at least 3% of the vote. Most election polls have a standard margin of error of 2-3 percentage points. We categorize average deviations as follows: up to 1 percentage point is "very good" (green); up to 2 points is "good" (yellow); up to 3 points is "acceptable" (orange), though this may indicate significant outliers for specific parties. Deviations exceeding 3 percentage points suggest a lack of predictive power and are rated "poor" (red).

Party deviation measures how frequently an institute’s data significantly diverges from the overall trend. Persistent outliers for specific parties can suggest methodological bias or "house effects." We analyze an institute’s deviation by comparing its party values against the PolitPro Election Trend—a weighted average of all current surveys. If a value is within +/- 1 percentage point of the trend, it is classified as "consistent." Beyond that, it is "higher" or "lower." We consider deviations in up to 20% of cases as normal. If deviations exceed 20%, we flag the party with a corresponding arrow. Note: Party deviation should always be analyzed alongside overall election accuracy.