CONFIDENCE.NORM Function (LibreOffice Calc)

• Introduced: LibreOffice 4.2.
• Availability: Supported in LibreOffice 26.8.1.1, verified against its function registry. This identifies a supported release, not a final working version.
• Scope: Desktop Calc, using the English function name. Availability in other spreadsheet applications is separate.
• Checked: 2026-10-14 — function reference; versioned support. Documentation/source check; not a new native test. Existing example-specific test claims below retain their stated scope.

CONFIDENCE.NORM calculates z(1−alpha/2)×sigma/SQRT(n). Sigma is the known population standard deviation, while n is the sample size. Under the normal-population and independent-random-sample assumptions, subtracting and adding this margin to the sample mean gives the corresponding confidence interval.

If sigma is replaced by an estimated sample SD, the same z calculation is an approximation rather than the exact unknown-sigma normal-population method. A Student t interval accounts for estimating sigma. Sample size alone is not a universal rule for choosing between the two.

=CONFIDENCE.NORM(Alpha; StDev; Size)

All three arguments are required. Alpha=0.05 requests 95% confidence, because the retained confidence level is 1−Alpha. StDev is the known population SD in the same units as the observations. Size is the number of sampled observations. Official function Help.

Although the Help labels Size as population size, the mean-interval formula uses sample size, not the number of people or items in the entire population. The release implementation divides by SQRT(Size) and floors a fractional Size before checking it.

These independently calculated values are rounded to twelve decimal places. The examples check mathematics and the cited contract, not a new native Calc run.

Suppose a sample mean is 50, the externally known population SD is 4, and there are 25 sampled observations. Put 0.05, 4, 25 and 50 in A2:D2. In E2 enter =CONFIDENCE.NORM(A2;B2;C2). Then use =D2-E2 and =D2+E2 for the bounds.

The margin is approximately 1.567971187632 and the interval is approximately 48.432028812368 to 51.567971187632. These numbers do not show that the sample was random or the supplied SD was appropriate; those assumptions come from the study design.

Keep the mean and SD in matching measurement units. Multiplying all measurements and sigma by ten also multiplies the margin by ten. A percentage-formatted alpha still supplies a proportion, so enter 5% rather than the number 5 for this example.

Increasing n from 25 to 100 with sigma unchanged halves the margin to 0.783985593816. The square-root relationship means that doubling sample size does not halve the margin. Higher confidence, obtained with smaller Alpha, increases the margin.

When sigma is unknown and estimated by STDEV.S, use CONFIDENCE.T for the usual normal-population t method. Use the same numeric observations for the mean, SD and count; a filtered SD and an unfiltered count do not describe the same sample.

A confidence interval for a proportion is a separate modeling choice. Substituting SQRT(p×(1−p)) gives a basic normal approximation that can behave poorly near 0 or 1 and in small samples. Do not treat that substitution as a universal proportion-interval method.

The cited implementation rejects Alpha outside the open interval (0,1), nonpositive StDev and a floored Size below 1. A genuine sample size should be an integer before calling the function. Import and locale coercion are separate from these numeric range rules.

Do not discard legitimate outliers merely to reduce the SD or narrow an interval. Correct demonstrated recording errors and document justified exclusions. Dependence, selection bias or inappropriate exclusions are not repaired by a formula that produces a small margin.

A 95% confidence procedure describes repeated-sample interval coverage for the population mean. It does not claim that 95% of individual measurements or future outcomes lie between these endpoints. Report the method and assumptions, not only the two numbers.

• Official CONFIDENCE.NORM syntax
• LibreOffice 26.8.1.1 calculation and numeric guards

No new native-application results, all-locale compatibility or extreme-tail precision are claimed.