A new election poll can attract attention because it looks different from the previous one. Sometimes that difference reflects real movement. Sometimes it is simply the normal variation that comes from interviewing a sample. Polling averages are designed to make that distinction easier to see.
What an average can do
Combining several polls reduces the influence of any one unusually high or low result. When different pollsters are sampling the same electorate at roughly the same time, an average can provide a steadier view of the underlying trend than a single survey.
The American Association for Public Opinion Research notes that averages can reduce the focus on individual noisy results and may have less error than individual polls because some survey errors vary from one poll to another.
What an average cannot do
An average does not eliminate bias shared by many polls. If the available surveys are systematically missing a group of voters, using similar turnout assumptions or reacting slowly to a genuine change in opinion, combining them will not automatically solve the problem.
An average is also not the same thing as an election forecast. A polling average describes the polls included in the calculation. A forecast attempts to estimate a future election outcome and usually requires additional assumptions or model inputs.
Recency and transparency matter
Older polls can become less informative when opinion is changing, while very small or methodologically opaque surveys can add noise. A useful polling tracker should therefore show readers which polls are included, their fieldwork dates, sample populations and pollsters.
How to use Votendi charts
Treat the average as a trend line, not a promise. Look at the direction and persistence of movement, then inspect the individual polls underneath it. If several independent polls move in the same direction, confidence that something has changed is stronger than when a single survey breaks away from the pack.
Votendi will keep the underlying polls visible because an average should be auditable. Readers should be able to move from the headline trend back to the surveys that produced it.
Sources
- AAPOR – Polling Accuracy and polling averages: https://aapor.org/polling-accuracy/
- Pew Research Center – Public Opinion Polling Basics: https://www.pewresearch.org/course/public-opinion-polling-basics/
- Pew Research Center – Key things to know about election polling: https://www.pewresearch.org/short-reads/2024/08/28/key-things-to-know-about-us-election-polling-in-2024/
Editorial note: This copy deliberately avoids specifying a proprietary weighting formula until Votendi’s final average methodology is documented.
