The width of a confidence interval is the distance between its upper and lower bounds. It equals twice the margin of error. If a 95% confidence interval for an average is 50 ± 4, the lower bound is 46, the upper bound is 54, and the width is 8.
Width tells you how precise your estimate is. A narrow interval means you have a tight, useful estimate of the true value. A wide interval means there’s a lot of uncertainty, and the true value could fall anywhere across a broad range.
How Width Is Calculated
A confidence interval follows a simple structure: your sample statistic (like a mean or proportion) plus or minus a margin of error. The margin of error itself has two components: a multiplier based on your confidence level, and the standard error of your estimate.
Since the interval extends equally in both directions from the center, the total width is:
Width = 2 × multiplier × standard error
The standard error depends on two things: how spread out your data is (the standard deviation) and how many observations you collected (the sample size). Specifically, standard error = standard deviation ÷ the square root of the sample size. So for a 95% confidence interval, where the multiplier is 1.96, the full width works out to 2 × 1.96 × (standard deviation ÷ √sample size).
Here’s a quick example. Suppose you measure the blood pressure of 100 people and find a mean of 120 with a standard deviation of 15. The standard error is 15 ÷ √100 = 1.5. For a 95% confidence interval, the margin of error is 1.96 × 1.5 = 2.94. The interval runs from 117.06 to 122.94, and the width is 5.88.
Three Factors That Control Width
Confidence Level
The confidence level you choose directly changes the multiplier in the formula, which changes the width. At 90% confidence, the multiplier is 1.645. At 95%, it’s 1.96. At 99%, it jumps to 2.576. A 99% interval from the same data will always be wider than a 95% interval, because casting a wider net gives you more certainty that you’ve captured the true value. The tradeoff is precision: you gain confidence but lose specificity about where the true value actually sits.
Sample Size
Increasing your sample size shrinks the standard error, which directly narrows the interval. Because sample size sits under a square root in the formula, the relationship isn’t one-to-one. Doubling your sample size doesn’t cut the width in half. It reduces it by about 29%. To actually halve the width, you need to quadruple the sample size. This diminishing return is why researchers carefully plan sample sizes before collecting data, balancing the cost of more observations against the precision they need.
Data Variability
When your data points are more spread out (higher standard deviation), the standard error increases and the interval gets wider. This makes intuitive sense: if you’re measuring something that varies wildly from person to person, any single sample gives you a less reliable picture of the whole population. You can’t control this factor the way you can control sample size or confidence level, but you can sometimes reduce variability through better measurement techniques or by studying a more homogeneous group.
What Narrow and Wide Intervals Tell You
A narrow confidence interval lets you make definitive statements about your estimate. If a study finds that a treatment lowers blood pressure by 8 points with a 95% confidence interval of 6 to 10, you can be fairly certain the real effect is somewhere in that tight range. That’s actionable information.
A wide interval tells you the opposite. If that same treatment showed a confidence interval of 1 to 15, you’d know the treatment probably does something, but you can’t say much about how much. The true effect might be trivially small or impressively large. Wide intervals typically signal that the study had too few participants, too much variability in the data, or both.
In medical research, where the 95% confidence level is standard, width often determines whether a study’s findings are considered clinically meaningful. Two studies might report the same average treatment effect, but the one with the narrower interval carries more weight because its estimate is more precise. When you see confidence intervals reported in a study, check the width before fixating on the point estimate. The width is where the real story about reliability lives.
Comparing Common Confidence Levels
To see how confidence level alone affects width, consider the same dataset analyzed at three levels. Using the multipliers for each:
- 90% confidence: multiplier of 1.645, producing the narrowest interval
- 95% confidence: multiplier of 1.96, about 19% wider than the 90% interval
- 99% confidence: multiplier of 2.576, about 57% wider than the 90% interval
The 95% level dominates in practice because it strikes a workable balance. A 90% interval is more precise but misses the true value 10% of the time. A 99% interval almost always contains the true value but can be so wide that it’s not very informative. The 95% level has become the default convention in most scientific fields, though there’s nothing magical about it. The right choice depends on how costly it would be to get the answer wrong.