CONFIDENCE.T 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.T uses the sample standard deviation because the population standard deviation is unknown. For an independent random sample from a normal population, the usual two-sided interval is sample mean ± t(1−alpha/2;n−1)×s/SQRT(n).

A small sample alone does not make this method correct for any data. Strong skew, dependence, biased sampling and data selected after inspection can affect coverage. Conversely, a large sample does not require switching from t to z when sigma remains unknown.

=CONFIDENCE.T(Alpha; StDev; Size)

All three arguments are required. Alpha is the total tail probability: 0.05 means a 95% confidence procedure. StDev is the sample SD, ordinarily calculated with STDEV.S. Size is the number of sampled numeric observations. The output is one margin, not two endpoints. Official function Help.

The Help’s generic labels refer to population SD and population size. For the Student t mean interval, use sample SD and sample size as in the statistical formula; the implementation evaluates a t critical value with Size−1 degrees of freedom.

The following margins were independently calculated and rounded to twelve decimal places; they are not newly recorded Calc runtime outputs.

Enter 10, 12, 14, 16 and 18 in A2:A6. Put these formulas in separate summary cells:

=AVERAGE(A2:A6)
=STDEV.S(A2:A6)
=COUNT(A2:A6)
=CONFIDENCE.T(0.05;STDEV.S(A2:A6);COUNT(A2:A6))

The mean is 14, sample SD is approximately 3.162277660168, and n is 5. The 95% margin is approximately 3.926486322955. Therefore the interval is approximately 10.073513677045 to 17.926486322955, assuming the sampling and distribution conditions hold.

Use the same observations in AVERAGE, STDEV.S and COUNT. A count from a larger range paired with an SD from a filtered subset would artificially change the margin. Keep calculations at full precision and round only the final displayed bounds.

The pinned release floors Size. It rejects Alpha≤0 or Alpha≥1, StDev≤0 and a floored Size below 1 as invalid arguments. A floored Size of exactly 1 instead produces a DivisionByZero error, because a t interval would have zero degrees of freedom. This is a source-confirmed distinction, not a native error screenshot.

For example, =CONFIDENCE.T(0.05;1;1) follows that division-by-zero branch. Fractional sample counts are not meaningful observations even when the function accepts and floors them. A zero sample SD also fails the positive-SD guard; do not silently turn every calculation error into a zero margin.

Do not remove outliers solely to obtain a narrower interval. Investigate recording errors and define exclusions using defensible, preferably prespecified criteria. Removing legitimate extreme observations changes the target population and the coverage claim.

Missing and erroneous cells also require an explicit data-cleaning decision. An error-filtering formula may make a worksheet calculate without making the remaining sample representative. Report the sample size, method, confidence level and any exclusions with the interval.

A confidence interval concerns the population mean under its sampling procedure. It is not an interval containing 95% of future individual observations. When the population SD is genuinely known, CONFIDENCE.NORM is the corresponding normal-critical-value calculation.

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

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