RSE Calculator — Relative Standard Error

Compute the relative standard error of an estimate as RSE = standard error ÷ mean × 100%. Enter the standard error directly, or provide a standard deviation and sample size to have the standard error worked out for you.

Quick Facts

Formula
RSE = (SE ÷ mean) × 100%, with SE = SD ÷ √n
A scale-free percentage: the smaller the RSE, the more precise the estimate.

Your Results

Calculated
Relative standard error
-
RSE = SE ÷ mean × 100%
Standard error (SE)
-
SD ÷ √n (or your entered SE)
Reliability band
-
Common agency thresholds

Ready

Enter the mean plus either a standard error, or a standard deviation and sample size.

Using the Relative Standard Error Calculator

The relative standard error (RSE), also called the coefficient of variation of the mean, expresses the standard error of an estimate as a percentage of the estimate itself. Because it is scale-free, it lets you compare the precision of estimates measured in completely different units — dollars, kilograms, counts of people — on a single 0-to-100% scale. A small RSE means the sampling error is tiny relative to the size of the estimate; a large RSE means the estimate could easily be off by a wide margin.

The formula

The relative standard error is defined as:

RSE = (SE / x̄) × 100%

where x̄ is the estimate (the sample mean) and SE is its standard error. When you have summary statistics from a simple random sample, the standard error of the mean is:

SE = SD / √n

so the RSE can be written directly as RSE = SD / (√n × x̄) × 100%, where SD is the sample standard deviation and n is the number of observations. This calculator lets you enter the standard error directly (useful when it comes from a survey report or a regression output), or supply the standard deviation and sample size and have SE = SD/√n computed for you.

Why relative standard error is useful

A raw standard error is hard to interpret on its own: an SE of 5 is negligible for an estimate of 5,000 but catastrophic for an estimate of 8. Dividing by the estimate removes the units and the scale, so an RSE of 2% always means the same thing — the standard error is 2% of the estimate — regardless of what is being measured. Statistical agencies such as the US Census Bureau, the Australian Bureau of Statistics, and many health surveys attach an RSE to every published estimate precisely so users can judge reliability at a glance.

Common reliability thresholds

There is no universal cutoff, but widely used conventions are:

  • RSE below 25%: the estimate is generally considered reliable and safe to publish.
  • RSE between 25% and 50%: the estimate is imprecise; it is often flagged with a warning and should be used with caution.
  • RSE of 50% or more: the estimate is too unreliable for most purposes and is frequently suppressed rather than published.

Different agencies set their own boundaries (some flag at 30%, some suppress at 33%), so always check the reliability standard that applies to your data source.

A worked example

Suppose a survey of n = 25 households produces a mean weekly spend of x̄ = $50 with a standard deviation of SD = $8. The standard error is SE = 8 / √25 = 8 / 5 = 1.6. The relative standard error is RSE = (1.6 / 50) × 100% = 3.2%. That is well under 25%, so the estimate is considered reliable. If the same standard deviation came from only n = 4 households, SE = 8 / 2 = 4, giving RSE = (4 / 50) × 100% = 8% — still acceptable, but noticeably less precise because the smaller sample carries more sampling error.

Frequently Asked Questions

What is the difference between RSE and the coefficient of variation?
They use the same idea of a ratio to the mean but at different levels. The coefficient of variation (CV) is the standard deviation divided by the mean (SD / x̄), describing the spread of the raw data. The relative standard error divides the standard error by the mean (SE / x̄), describing the precision of the sample mean. Since SE = SD / √n, the RSE equals the CV divided by √n, so RSE is always the smaller of the two for n greater than 1.
Should I enter the standard deviation or the standard error?
Enter whichever you have. If you already know the standard error — for example, from a published survey table, a confidence interval, or regression output — type it into the standard-error field and it will be used directly. If you only have the raw-data standard deviation and the sample size, leave the standard-error field blank and the calculator will compute SE = SD / √n for you before finding the RSE.
Why must the mean be non-zero?
Relative standard error divides by the estimate, so it is undefined when the mean is zero and unstable when the mean is close to zero. If your estimate is near zero, a tiny standard error can still produce a huge or infinite RSE, which is why RSE is not a meaningful precision measure for quantities that can legitimately be zero or that change sign.

Related calculators