FTEST Function (Calligra Sheets)

• Introduced: First release not established; the name is present in Calligra 2.4.0 source.
• Supported in: Calligra Sheets 26.08.2, with no retirement established.
• Scope: Desktop Sheets with the Statistical module. Compatibility names remain registered; no retirement is inferred. Examples use English semicolon-separated formulas.
• Checked: October 14, 2026. Selected formulas ran through the official 25.04.2 engine library. Source and independent mathematics were checked. This is engine evidence, with no GUI-cell or import verification; 26.08.2 was source-checked only.

FTEST calculates sample variances, puts the larger variance in the numerator, and doubles that F distribution’s upper tail. It does not take the smaller of the two tails. For unequal sample sizes, that shortcut can exceed 1 and fail the ordinary two-sided p-value interpretation. Function-specific official reference.

=FTEST(array1;array2)

Exactly two numeric sample arrays are required. Each must contain at least two numeric values and have nonzero sample variance. The callback independently skips nonnumeric entries, including the error values tested here; it does not enforce paired observations. Registration and implementation.

The engine and mathematical columns are separate. Results are rounded to twelve decimal places, or thirteen significant digits in scientific notation for very small or large values. This is a display choice, not an accuracy guarantee. Inline arrays in braces are formula values, not a workbook-import test. Their semicolons separate columns and vertical bars separate rows. Strings remain explicitly quoted, and booleans use TRUE()/FALSE().

For {1;2;3} and {2;4;6}, sample variances are 1 and 4. The ratio is 4 with degrees 2 and 2, and the doubled upper-tail result is 0.4 in either sample order.

The discriminatory case is {-1;0;1} versus {0;0;0;2}. Both sample variances are exactly 1, but sizes are 3 and 4. On a tie, the source chooses the second sample as numerator. The resulting degree order changes when the arrays are swapped. The table’s mathematical column uses 2×min(lower tail,upper tail) for the pre-specified variance ratio; it remains within [0,1].

For these equal-variance samples, a ratio fixed as sample A variance divided by sample B variance is 1, with degrees 2 and 3. The executed explicit calculation =2*MIN(LEGACYFDIST(1;2;3);1-LEGACYFDIST(1;2;3)) returned approximately 0.929516003090. This verifies the ordinary two-tail arithmetic for this example; it is not permission to apply an F variance test without its independent-sample normal-model assumptions.

Clipping the faulty output to 1 would hide the problem and would not reproduce the two-tail result. Preserve variance orientation and degrees together. The additional saved symmetry check uses the simpler 1-versus-4 variance example, where the shortcut happens to work.

The executed advanced check was:

=FTEST({1;2;3};{2;4;6})-FTEST({2;4;6};{1;2;3})

Its engine result was 0.

Single-value samples and constant samples returned #N/A. Numeric text, booleans and an embedded division error were skipped by the callback’s numeric-only loop. That can conceal an incomplete or damaged sample, so check counts and input errors before using the calculation.

The result-above-one warning is tied to the stated engine and exact sample arrays. The matching current callback documents the same algorithm; it does not constitute a newer-version execution test.

• Function-specific implementation
• Versioned implementation reference
• Legacy F Cdf helper
• Beta helper
• Array Parser helper

The statistical and math modules and value-calculation helpers are byte-identical between the inspected 25.04.2 and 26.08.2 sources. That comparison supports the source discussion but does not establish later-version execution, GUI-cell behavior or other locales.

VAR, LEGACYFDIST, FDIST.