STDEVPA Function (Gnumeric)
Learn the STDEVPA function in Gnumeric: Returns the standard deviation based on the entire population. Text is evaluated as zero.
Verification status: Official documentation referenced; example outcomes independently reasoned or arithmetically checked where stated. Not executed in Gnumeric. How formula examples are checked
Native-validation limits: All expected outputs are independently calculated; no Gnumeric execution or screenshots were performed.
Quick answer
=STDEVPA(A2:A7) estimates the population standard deviation of a selection of worksheet cells. Unlike =STDEVP(A2:A7), it includes text as zero and Boolean values as 1 (TRUE) or 0 (FALSE) while ignoring empty cells. The official Gnumeric Statistics manual documents these rules for STDEVPA; they are not guesses based on another spreadsheet. A practical use is measuring dispersion in a deliberately mixed-value dataset, provided treating words and flags as numeric observations makes sense. Official function definition
Choose this form if your data genuinely consists of the entire population to be described. For data where text labels and Boolean flags should not count as measurements, use the numeric-only variant STDEVP instead. The result has the same unit as the original numeric measurements. Unlike the variance, it is the square root of the mean squared dispersion.
Gnumeric syntax and parameters
=STDEVPA(area1,area2,…)
| Parameter | Meaning | Required/default |
|---|---|---|
area1 |
First cell area or range holding observations, such as A2:A7 |
Required; the manual documents no fallback default. |
area2, … |
Additional cell areas, such as B2:B3, included in the same calculation |
Optional subsequent areas; no numeric defaults are documented. |
The documented argument order is first area, then additional areas; there is no separate population or ignore_text switch. For an ordinary single-area formula use =STDEVPA(A2:A7); for noncontiguous areas use =STDEVPA(A2:A7,B2:B3). Areas are not weights: every included data value contributes one observation. Official synopsis
The population standard deviation calculation uses denominator n after documented coercion, where n counts all nonempty numeric, text and logical entries in the selected areas. For text cells, the value is 0; for genuine Boolean TRUE and FALSE values, the values are 1 and 0; empty cells are omitted, not counted as zero. A cell containing the text "TRUE" is a string, not a Boolean: it is counted as 0, not 1. Gnumeric data-type rules
Version, edition and locale: This research is scoped to Gnumeric 1.12.62, with the official Gnumeric 1.12 Statistics manual checked 2026-10-13. The public catalogue lists the function as a complete implementation with basic upstream testing, but catalogue registration is not a runtime test on your installed OS, Gnumeric build, plug-ins or spreadsheet. Examples use English function names, commas as argument separators and periods as decimal points. Gnumeric draws language/number formatting settings from its localization environment; verify the separator and formatting on your actual system rather than blindly replacing characters. The documentation does not state a numeric default for omitted areas. 1.12.62 announcement · Catalogue · Localization
Reproduce the example in your own worksheet
Enter exactly these six cells. For A4 type the letters missing (without quote marks), and for A7 leave the cell truly empty. In A5 and A6 enter formulas so that the results are actual logical/Boolean values rather than string labels.
| Cell | Enter | Underlying interpretation for STDEVPA |
|---|---|---|
| A2 | 2 |
Number 2 |
| A3 | 4 |
Number 4 |
| A4 | missing |
Text → 0 |
| A5 | =1=1 |
Boolean TRUE → 1 |
| A6 | =1=2 |
Boolean FALSE → 0 |
| A7 | (leave blank) | Ignored (not an observation) |
Then enter =STDEVPA(A2:A7) in an unused cell. Expected result (independently calculated, not observed in Gnumeric): 1.4966629547. The five counted values are 2, 4, 0, 1, 0, so n = 5, their sum is 7, their mean is 7/5 = 1.4, and the sum of squared deviations from that mean is 11.2. Therefore STDEVPA uses √(11.2 ÷ 5) = √2.24 ≈ 1.4966629547. In a narrow column, Gnumeric may display fewer decimal places than the internally computed value. Formula and cell-type semantics
To see why the A suffix matters, put =STDEVP(A2:A7) in a different cell. Based on the numeric-only function’s documented purpose, only the numeric entries 2 and 4 are the measurements; the independent mathematical comparison is 1. This differs because STDEVPA deliberately counts A4, A5 and A6. This contrast is particularly useful when diagnosing imported spreadsheets with text placeholders. Related numeric-only function
Advanced examples
Noncontiguous areas / incremental data
Enter 6 in B2 and 8 in B3, then calculate:
=STDEVPA(A2:A7,B2:B3)
The combined seven counted values are 2, 4, 0, 1, 0, 6, 8; their sum is 21, mean is 3, and squared-deviation sum is 58. Expected result (independently calculated): 2.8784916685. If you used only A2:A7, you would exclude B2:B3 entirely. The comma separates areas here; it is not a decimal separator. Multi-area syntax
Cross-check dispersion and audit unintended zeroes
The corresponding pair of functions should agree mathematically: VARA is the square of STDEVA, and VARPA is the square of STDEVPA, within floating-point rounding. For these inputs, independently calculated values are sample variance 2.8, population variance 2.24, sample standard deviation 1.6733200531, and population standard deviation 1.4966629547. Thus you can compare =STDEVPA(A2:A7) with the paired VARPA guide. These comparisons are mathematical checks, not claimed Gnumeric execution results. Official sample/population definitions
Decision check: When missing really denotes unknown measurement rather than zero measurement, treating it as 0 distorts dispersion. In that case, clean or filter the data, use genuinely blank cells for omissions, and consider =STDEVP(A2:A7) if the numeric-only interpretation matches your analysis. Do not replace missing measurements with zeros simply to make this formula return a number.
Errors, edge cases and troubleshooting
| Symptom or situation | Explanation and action | Confidence |
|---|---|---|
A4 (missing) unexpectedly changes the result |
Text is an included zero in STDEVPA. Replace unknown placeholders with genuinely blank cells if you intend omission; or choose STDEVP when numeric-only behavior is appropriate. |
Explicit official text/Boolean rule. |
| A5 or A6 behaves like text rather than a logical value | Check that A5 contains =1=1 and A6 =1=2; typing the words TRUE or FALSE into imported text fields is not a reliable type test. |
Official Gnumeric text/Boolean distinction; data entry must be checked locally. |
| Deleting a value changes the sample size | Empty cells are ignored. A space or a formula returning an empty-looking text string is not necessarily an empty cell; inspect the underlying cell type before interpreting it. | Blank-cell rule documented; formula-empty-string edge case not natively tested here. |
| Dispersion cannot be estimated from too few counted values | If there are no counted observations, the population denominator n is zero and the statistic is mathematically undefined. Add enough meaningful observations before interpreting the statistic. This guide deliberately does not assign an exact Gnumeric error code without native execution or a release-matched implementation test. | Mathematical constraint; exact cell error pending native test. |
| The result differs slightly in the last decimals | Floating-point computation or display formatting can introduce small differences. Compare rounded values and retain sufficient precision; do not invent a software-specific numerical guarantee. | General numerical limitation; native precision untested. |
| Formula is rejected or prints unexpected separators | Confirm a valid range, argument punctuation and your current locale, and ensure the cells are not plain text where formulas are expected. | Official syntax and locale documentation; local execution needed. |
A source cell contains an error (#VALUE!, #N/A, etc.) |
Inspect that source first and repair or explicitly handle it; the official STDEVPA manual excerpt does not specify every propagated error code, so the exact result remains to be verified in Gnumeric. | Limitation stated rather than guessed. |
Compatibility and version caveats
Gnumeric’s official manual marks STDEVPA as Excel compatible, which is useful context, but it does not justify copying every Microsoft Excel edge case, error code, imported-file conversion rule or locale assumption into a Gnumeric claim. This guide’s worked examples are based on Gnumeric’s documented counting semantics and independent arithmetic. OpenDocument/LibreOffice compatibility is not asserted; both export format and target-app formula behavior require separate checks. This review uses official Gnumeric 1.12.62 source; older installed releases, optional plugins, localization settings and future releases may differ. Gnumeric Statistics reference · 1.12.62 announcement
Related functions
Primary source verification: The official 1.12.62 archive was checksum-verified; the statistical callback confirms range text-to-zero, Boolean-to-0/1 and blank omission, with separate sample and population numeric routines. Direct-argument coercion is not inferred from range rules.
Official sources, evidence and validation
Sources checked 2026-10-13 (UTC): the Gnumeric Statistics manual STDEVPA synopsis, arguments, description and documented Excel compatibility; the Gnumeric function catalogue (function registration / upstream completeness classification, not direct runtime proof); the Gnumeric 1.12.62 release announcement (release scope); and the Gnumeric localization manual (locale-specific configuration). Each normative claim above links to its specific official page. Documentation and independent arithmetic are distinct evidence types.
Validation: Official documentation and checksum-verified Gnumeric 1.12.62 statistical callbacks inspected; all sample/population and mixed-type examples independently recalculated. Native Gnumeric execution, locale regression and workbook round-trip testing were not performed.