CONFIDENCE.T in ONLYOFFICE Spreadsheet Editor

• Introduced: Original introduction release is not established by the sources checked.
• Available: Documented in current unversioned ONLYOFFICE function help, checked October 18, 2026.
• Implementation inspected: Official SDKJS v10.0.0.169, commit 3679e0375a758d7dc59d31337a45461be3525ed8. This establishes a source snapshot, not identical behavior in every installed editor.
• Retired or deprecated: Retirement or formal deprecation is not established by these sources. A legacy spelling is not automatically a removed function.
• Testing: Examples are independently calculated or labeled source predictions. No ONLYOFFICE application or calculation-engine execution was performed. No native screenshot or general import/export guarantee is claimed.

Use English names and comma separators as shown; adapt separators outside quoted text if your locale requires semicolons. Keep full precision in calculations and round only the displayed summary.

=CONFIDENCE.T(alpha, standard_dev, size)

Exactly three arguments are registered. The function does not accept a mean and does not return a pair of endpoints. Its output is the margin, or half-width, of the usual symmetric interval. Full interval width is twice that number.

The t-based calculation uses t(1−alpha/2,n−1) × s / sqrt(n). In the usual unknown-population-variance mean interval, s is the sample standard deviation and the t distribution has n−1 degrees of freedom. The calculation does not validate those assumptions from the three numbers supplied.

ONLYOFFICE help labels standard_dev as a population standard deviation, but the familiar t-interval interpretation uses the sample estimate. The source merely multiplies the supplied number by a t critical value; it cannot determine its origin. NIST’s t-interval explanation gives the sample-standard-deviation formulation.

Use a small arithmetic demonstration with observations 8, 10 and 12 in L2:L4. Their mean is 10. The squared deviations from 10 are 4, 0 and 4, so the sample variance is 8/(3−1)=4 and the sample standard deviation is 2.

1. Put alpha 0.05 in A2, standard deviation 2 in B2 and sample size 3 in C2.
2. Enter =CONFIDENCE.T(A2,B2,C2) in D2.
3. The independent half-width is approximately 4.96827542350.
4. With the sample mean 10 in E2, use =E2-D2 and =E2+D2 for endpoints 5.03172457650 and 14.96827542350.

This tiny dataset is for checking the formula mechanics, not evidence that any real sampling assumptions hold. Its two degrees of freedom give a 97.5th-percentile t value of about 4.30265272975. Multiplying by 2/sqrt(3) gives the half-width.

The df=2 t cumulative distribution has the closed form 0.5+t/(2*sqrt(t^2+2)). Setting it to 0.975 and solving gives the stated critical value, providing an independent mathematical check rather than relying on another spreadsheet’s t implementation.

A 95% two-sided interval uses alpha 0.05, not 0.95. The critical point is based on alpha/2 in each tail. Decreasing alpha asks for higher confidence and generally widens the interval; increasing the standard deviation also widens it.

The returned half-width has the same units as standard_dev. If the measurements are in seconds, the margin is in seconds, not a percentage. Formatting the margin as a percentage does not convert it into a confidence level.

For a t interval, increasing sample size changes both the square-root denominator and the critical value through df=n−1. Do not assume that quadrupling a small sample exactly halves the t margin. As a separate easy check, alpha 0.5, standard deviation 2 and n=2 gives t critical value 1 with df=1, so the margin is sqrt(2)=1.41421356237.

A confidence interval for a mean is not a prediction interval for one future observation. Nor does a computed 95% interval establish a 95% posterior probability for a fixed parameter without an appropriate probabilistic framework. The spreadsheet computes a standard formula; interpretation depends on the sampling model.

The implementation rejects alpha at or outside 0–1, standard_dev at or below zero and size below 1 after integer parsing, using the numeric-error category, conventionally #NUM!. The table’s error is a source prediction rather than an application observation. Size is parsed as an integer, so enter the actual whole sample count rather than relying on fractional truncation.

Although n=1 passes the initial common guard, a t reference would then have zero degrees of freedom. Use n>=2 and do not infer a valid interval from the initial size check. The function calls the shared bounded inverse solver. Bracket failure is flagged, but not every refinement-limit exit is converted to an error; extreme numerical behavior is not established here.

These are independent mathematical expectations or identified source-error predictions, not native application results. Preserve full input precision when reproducing them.

Use the exact probability direction, parameter order and data meaning above when moving a workbook between products. Similar names are not a general cross-version or file-format compatibility guarantee.

AVERAGE calculates a sample mean; COUNT counts numeric observations; SQRT exposes the standard-error calculation.

• Official CONFIDENCE.T help, checked October 18, 2026.
• Pinned SDKJS implementation, including parameter validation and numerical helpers.