Standard error is a measure of how precise a sample statistic is as an estimate of the true population value. If you surveyed 100 people to estimate the average height of all adults in a country, the standard error tells you how much that sample average is likely to bounce around from one 100-person sample to the next. A smaller standard error means your estimate is more reliable; a larger one means there’s more uncertainty.
The concept is foundational in statistics because it’s the key ingredient for building confidence intervals and running hypothesis tests. Nearly every time you see a margin of error or a p-value in a study, standard error is doing the work behind the scenes.
How Standard Error Differs From Standard Deviation
This is where most of the confusion lives. Standard deviation and standard error sound similar and are mathematically related, but they answer different questions.
Standard deviation describes how spread out individual data points are. If you measured the blood pressure of 200 people, the standard deviation tells you how much those 200 readings vary from person to person. It’s a snapshot of real-world variability in your sample.
Standard error, on the other hand, describes the uncertainty of a summary statistic like the average. It answers: if I repeated this study many times, how much would the average jump around? Think of standard deviation as describing the scatter of individual people, and standard error as describing the reliability of the group’s average.
A critical practical difference: as you increase your sample size, the standard deviation stays roughly the same (because the underlying variability in the population hasn’t changed), but the standard error shrinks. Collecting more data doesn’t make people less variable, but it does make your estimate of the average more precise.
The Formula
For the most common case, the standard error of the mean, the formula is straightforward:
SE = SD / √n
That’s the standard deviation of your sample divided by the square root of the sample size. So if your sample of 100 people has a standard deviation of 10, the standard error is 10 / √100 = 1.0. Quadruple the sample to 400 and the standard error drops to 10 / √400 = 0.5.
Notice the square root relationship. You don’t get a proportional payoff for adding more people. To cut the standard error in half, you need to quadruple your sample size, not double it. This tradeoff is one of the core considerations when researchers plan how large a study needs to be.
For proportions (like the percentage of people who prefer a product), the formula changes slightly. If the proportion is p and the sample size is n, the standard error is:
SE = √[p × (1 − p) / n]
This is what drives the margin of error you see in political polls. A poll of 1,000 people with a 50/50 split has a standard error of about 1.6 percentage points, which produces the familiar “plus or minus 3 points” margin when doubled for a 95% confidence interval.
How It Powers Confidence Intervals
The most visible use of standard error is in confidence intervals. A 95% confidence interval is typically calculated as the sample estimate plus or minus about two standard errors. If the average weight in your sample is 70 kg and the standard error is 2 kg, the 95% confidence interval runs from roughly 66 to 74 kg. That interval is your way of saying: “We’re fairly confident the true population average falls somewhere in this range.”
The standard error is what determines the width of that interval. A small standard error produces a tight, precise interval. A large one produces a wide, less informative interval. This is why sample size matters so much in research: larger samples yield smaller standard errors, which yield narrower confidence intervals, which give you more useful answers.
How It Powers Hypothesis Tests
When researchers run a t-test or z-test to determine whether a result is statistically significant, the standard error sits in the denominator of the test statistic. The basic structure looks like this:
test statistic = (sample statistic − hypothesized value) / standard error
In other words, the test asks: how far is my result from what I’d expect if nothing were going on, measured in units of standard error? If your sample mean is three or four standard errors away from the hypothesized value, that’s strong evidence the difference is real, not just noise. If it’s less than two standard errors away, the result is generally not considered statistically significant.
For tests involving proportions, the test statistic is called a z-score. For tests involving means, it’s called a t-statistic. The mechanics are the same: the observed difference divided by the standard error tells you whether the result is likely due to chance.
Reading Error Bars on Graphs
In scientific papers and presentations, you’ll often see bar charts or line graphs with thin lines extending above and below each data point. These “error bars” can represent standard deviation, standard error, or a 95% confidence interval, and which one matters a lot for interpretation.
Standard error bars are shorter than standard deviation bars (because SE is always smaller than SD for any sample larger than one). When comparing two groups, the degree of overlap between their SE bars gives you a rough sense of statistical significance. If you have 10 or more observations per group and there’s a visible gap between the SE bars, the p-value is approximately 0.05. If the gap is twice the length of one SE bar, the p-value drops to around 0.01.
The catch is that many graphs don’t clearly label which type of error bar they’re using. Research in cardiovascular journals found that 64% of articles misused or mislabeled the standard error. Many journals now recommend that authors explicitly state whether bars represent SD, SE, or confidence intervals, but not all do. When you see error bars, check the figure legend before drawing conclusions.
When Each Measure Is Appropriate
The choice between reporting standard deviation and standard error depends on what you’re trying to communicate. If the goal is to show how much variation exists in the raw data (for example, the range of patient ages in a study), standard deviation is the right choice. It’s a descriptive statistic that characterizes individual-level spread.
If the goal is to show how precisely you’ve estimated a population parameter, standard error is appropriate. It’s an inferential statistic that speaks to the quality of your estimate, not to individual variability. In practice, many researchers recommend reporting confidence intervals rather than standard error alone, since a confidence interval directly communicates the range of plausible values in a way that’s easier for most readers to interpret.
One common pitfall is using the standard error to make data look less variable than it actually is. Because SE is always smaller than SD, reporting “mean ± SE” produces a tighter-looking result than “mean ± SD.” This isn’t inherently wrong, but it can be misleading if readers assume the ± symbol represents variability among individuals rather than precision of the mean. The simplest fix is to always specify which measure you’re reporting.