Confidence Interval

For many common estimates, a confidence interval is built from three ingredients: the estimate itself, a measure of how much estimates would vary across repeated samples, and a critical value from a probability distribution. As sample size grows, that sampling variability usually shrinks, which is why confidence intervals often get narrower in larger studies. But narrowing also depends on noise in the data and on whether the model assumptions are reasonable.

The number that pretends to be exact

A paper says magnesium improved sleep score by 3.2 points. That looks crisp, almost surgical. Then you notice the line beside it: 95% CI, 0.4 to 6.0. Suddenly the result stops being a pin and starts looking more like a flashlight beam. That is the trap with confidence intervals: readers remember the center number and skim past the spread, even though the spread often tells the more honest story.

Why 95% does not mean “95% chance this study is right”

Picture an archer shooting at the same target again and again in gusty wind. Each arrow lands in a slightly different place. A study estimate works the same way: if you repeated the whole study many times, the estimate would wander because samples differ by luck alone. A confidence interval is the range produced by a method designed so that, over many repeats, 95% of those ranges would capture the true value when the method’s assumptions hold.

That is why the common phrase “there is a 95% chance the true value is inside this interval” is not quite right. After you calculate one interval, it either did or did not catch the truth; the 95% belongs to the procedure, not to your single finished interval.

What the notation is showing you

In papers, confidence interval notation often appears as estimate (95% CI 1.2 to 4.8), 95% CI [1.2, 4.8], or as horizontal bars on a confidence interval graph or forest plot. The center is the estimate. The ends are the lower and upper confidence limits. A confidence interval table usually lists the estimate, standard error, and the two limits. In many intro classes, the confidence interval formula for a mean is taught as:

estimate ± critical value × standard error

For a 95% confidence interval using the normal distribution, that critical value is often about 1.96, which is why people search for the 95% confidence interval z score. But that shortcut is not universal; different data types and smaller samples may use different distributions and assumptions.

Width matters more than the badge

The most useful question is usually not “Is it 95%?” but “How wide is it?” A narrow interval means the estimate is tightly pinned down. A wide one means the study is telling you, “somewhere in this neighborhood.” If the interval crosses the “no effect” point—often 0 for mean differences or 1 for ratios like risk ratios—the data are compatible with little or no effect as well as benefit or harm, depending on the measure.

One practical decision: when comparing two supplement studies, trust the one with the narrower interval around a meaningful effect, not just the one with the more exciting headline number. That single habit will make you a better reader of evidence immediately.