Why repeat the same question
A single poll gives a snapshot. It can tell you roughly how many people say they hold a view at one moment, but not whether that number is rising, falling or steady. To describe change, you need at least two measurements that can fairly be compared. The most dependable way to get them is to ask the same question, in the same way, of a comparable group, again and again.
That is what a tracking survey does. Some track opinion monthly, others yearly or at longer intervals. Because the method in good tracking surveys stays fixed, differences between waves can be attributed more plausibly to a real shift than to a change in how the data was gathered. Long-running series on topics such as satisfaction, trust or attitudes to social issues are among the most valued resources in social research for exactly this reason.
The logic of holding things constant
Every poll result mixes the thing you want to measure with the particulars of how it was measured: the wording, the answer options, the sample, the mode and the timing. If all of those stay fixed, then what changes between waves is more likely to be genuine change in the population. If they shift, the line on the chart may move for reasons that have nothing to do with opinion.
This is why survey organisations often keep a question word for word for decades, even when the phrasing starts to feel dated. Improving a question can make it better in itself while breaking its comparability with the past. That trade-off is real, and it is usually settled by running old and new versions side by side for a period.
Two ways to follow change
Repeated cross-sections
Most tracking surveys draw a fresh sample each time. Each wave describes the population at that moment, and the comparison across waves shows how the overall picture has changed. The strength is that every wave is independent and not affected by earlier participation. The limit is that it shows only net change. If one in ten people moves one way and another one in ten moves the other way, the headline stays flat and the churn is invisible.
Panels that re-interview the same people
A panel study follows the same individuals over time. This can show who changed their mind, in which direction, and whether certain kinds of people are more likely to move. It reveals movement that cross-sections average away. The costs are that people drop out over time, that repeated questioning can influence answers, and that the remaining group may gradually become less representative. Good panel studies work to reduce attrition and adjust for it.
What can break a trend line
Even carefully run series are vulnerable. The following are common causes of apparent change that is really a change in measurement.
- Wording changes. A small edit to a question or its introduction can shift answers, as explored under question wording. Even a change in the order of questions that came before can matter.
- Mode changes. A move from telephone interviews to online self-completion, which many long-running surveys have made, can alter responses, particularly on sensitive topics where people differ in how candid they are in each setting. This overlaps with social desirability.
- Sampling changes. Different lists, different recruitment, or a new panel provider can produce a different mix of people. Differences in sampling frames can change who is counted at all.
- Changing response patterns. If willingness to take part shifts, for instance after a period of heavy news coverage, the people who answer in one wave may differ from those in the last. This is the territory of non-response bias.
- Changing weights. Revised population benchmarks or a new weighting scheme can move headline figures even when the raw data are similar.
- Changing meaning. The words in a question can mean something different to people over time. A term describing a social group, a technology or a political label may not mean the same thing across decades.
Responsible survey organisations typically flag these breaks in a series, sometimes by drawing a visible gap in the chart or by publishing a bridging study that compares old and new methods in the same period.
Noise versus signal
Because every wave has sampling uncertainty, a series will wobble even if nothing real has changed. A move of a point or two from one wave to the next is often within the range of chance, and it can be misleading to explain it with the latest news event. Looking at a run of several waves, or at a smoothed average, is generally a better guide than reading each data point as a message. Understanding the margin of error attached to each wave helps in judging which movements are worth taking seriously.
It also helps to check whether other independent series show a similar pattern. If several trackers with different methods point the same way, the case for a real shift is stronger than if only one does.
Comparing different polls is not the same as tracking
A common mistake is to compare a figure from one pollster in one year with a figure from another pollster in another year and describe the gap as a trend. Different wording, samples and methods make this unreliable. Trend stories that stitch together unrelated data points can look convincing while establishing very little. Only a series with a consistent method can fairly show change, and even that tells you how opinion moved, not why. Linking a shift to a cause requires more evidence than a line on a chart, which is the territory of causal claims.
Practical habits
- Check that the question wording and answer options are identical across the waves being compared.
- Look for notes about changes in method, and be cautious around those dates.
- Read several waves, not just the latest pair.
- Ask whether the series follows the same people or fresh samples, since this affects what the change means.
- Treat explanations for a shift as hypotheses unless there is evidence beyond the trend itself.
Repeated questions are not a perfect instrument, but they provide the clearest view available of whether opinion is moving, which is why they anchor so much of what is known about changing attitudes.