What Are Confidence Intervals and Margin of Error?
When you see a research result, you typically focus on a single number: '62% of consumers liked the product.' But beside that number sit two often-ignored concepts: confidence interval and margin of error. These two tell you how reliable that 62% actually is. In this article, we explain what they mean, how they're calculated, and why they matter when reading results.
What is margin of error?
Margin of error shows how much a research result can deviate from the true value. Since you conduct research on a sample rather than the entire population, your result is an estimate close to the true value, not the exact value itself.
Say 62% of participants liked a product and the margin of error is 3%. This means the true approval rate is most likely somewhere between 59% and 65%.
So 62% should be read not alone but with the band around it. As margin of error shrinks, this band narrows and your estimate sharpens.
What is a confidence interval?
A confidence interval is the margin of error applied to the result. In the example above, the band between 59% and 65% is the confidence interval itself.
Margin of error expresses the magnitude of deviation; confidence interval expresses the concrete number range that deviation creates. They are two sides of the same coin.
What makes a confidence interval meaningful is the confidence level that accompanies it. Because saying 'the true value is between 59-65%' isn't enough on its own. You also need to state how sure you are.
Where does confidence level fit in?
Confidence level expresses how sure you are that the confidence interval is correct. The standard in research is 95% confidence level.
Confidence level
Answers 'how sure am I.' 95% confidence: if you repeated the study 100 times, the interval would contain the true value in about 95 of those attempts.
Margin of error
Answers 'how much can it deviate.' The limit of how far the result can stray from the true value.
Confidence interval
Answers 'what range is the true value in.' The concrete number range created by applying the margin of error to the result.
Raising confidence level (e.g., to 99%) provides more safety but at a cost: either margin of error grows or you need more participants.
How is margin of error calculated?
Here's what the formula looks like:
Margin of Error = Z × √(p(1-p) / n)
- Z = Value corresponding to confidence level (1.96 for 95%)
- p = Observed proportion
- n = Sample size
You don't need to solve this by hand every time. But the formula tells you something important: margin of error depends on the square root of sample size.
The practical implication: to halve the margin of error, you need to quadruple—not double—the sample.
How does sample size affect margin of error?
The table below shows approximate margins of error for different sample sizes at 95% confidence level (worst-case p=0.5 assumption):
| Sample size | Approximate margin of error |
|---|---|
| 100 | 9.8% |
| 200 | 6.9% |
| 400 | 4.9% |
| 600 | 4.0% |
| 1,000 | 3.1% |
| 2,000 | 2.2% |
With 100 people, margin of error is ~10%—useless for most business decisions. At 600 it drops to 4%. Going from 1,000 to 2,000 only reduces it by about one point. That's why 'what's my acceptable margin of error' is the starting point for sample planning.
Common mistakes when analyzing results
Not using confidence intervals and margin of error correctly leads to misreading research results:
Not seeing margin of error at all
The difference between 62% and 58% may not be statistically significant if margin of error is 4%. Both numbers may actually be within the same range.
Forgetting subgroup margin of error
Even with 1,000 total sample, an 80-person subgroup has much higher margin of error. Carrying the precision of total results to subgroups is misleading.
Presenting small differences as big news
If a score went from 61% to 63% and margin of error is 4%, there's nothing to celebrate. The difference must exceed margin of error to indicate real change.
Ignoring confidence level
95% and 90% confidence levels produce different ranges. When comparing two studies, ensure they use the same confidence level.
Does Sorbunu handle this calculation for you?
Yes. When you create a research project on Sorbunu, your targeted confidence level and margin of error are automatically calculated alongside sample size.
You can see how margin of error changes as you adjust parameters before even going to field, and our research team guides you accordingly.
When results arrive, reports present findings alongside confidence level and margin of error information. Especially in subgroup comparisons, seeing whether a difference is statistically significant prevents decisions based on wrong conclusions.
Because the real risk isn't wrong data—it's misreading correct data.
Frequently Asked Questions
No, but they're closely related. Margin of error is a single value showing how much the result can deviate (e.g., 3%). Confidence interval is that deviation applied to the result—a concrete number range (e.g., 59-65%).
Imagine repeating the same study under the same conditions 100 times. 95% confidence level means the confidence interval you found would contain the true value in about 95 of those attempts.
The most direct way is increasing sample size, but due to the square root relationship, halving the margin of error requires quadrupling the sample. Beyond a certain point, setting an acceptable threshold rather than growing infinitely is more practical.
Simple check: if the difference between two results is larger than the margin of error, the difference is likely real. If it falls within the margin of error, the two results are considered statistically indistinguishable.
Because estimates from few people are inherently less precise. In a 100-person sample, a few people answering differently can significantly change the result. As sample grows, individual fluctuations' impact decreases and the interval narrows.
No. Confidence intervals are calculated for means, scores, and other measures too. For example, a satisfaction average or NPS score also has a confidence interval. The logic is the same: the obtained value is an estimate with an uncertainty band around it.
Related Solutions
Brand Awareness & Perception Research
Measure awareness and perception scores with reliable margins of error and track over time.
Customer Satisfaction Research
Monitor satisfaction scores in a statistically meaningful way.
Product & Concept Testing
Clarify whether differences between alternatives are real or noise.
Read Your Results with Confidence
On Sorbunu, research results are presented with confidence interval and margin of error information. Read correct data correctly.