DSTDEV in Collabora Office: Sample Standard Deviation
Measure sample standard deviation for matching records, audit the number of numeric observations, and distinguish a constant sample from missing data.
Verification status: See the evidence and test scope described in this guide. How formula examples are checked
Quick answer: DSTDEV calculates sample standard deviation for numeric values in a field of matching database records. For 4, 8 and 12 minutes, the expected result is 4 minutes. The result uses the original measurement unit, unlike variance.
Availability and version history
This guide is based on the official DSTDEV reference retrieved October 22, 2026. The saved page is labeled Collabora Office 26.04 Help, but the /latest/ URL can change. It does not prove the first release, current free-edition entitlement, installed-build behavior, or Online/mobile parity. Numeric examples are independently checked, not captured Collabora outputs.
Syntax and the selected sample
=DSTDEV(Database; DatabaseField; SearchCriteria)
Supply a database including headers, the field to calculate, and a separate header-and-condition range. All three arguments are required. For the calculation field, use a quoted header, a reference to its header cell or a one-based position within the database.
Criteria choose the records first. DSTDEV then uses numeric values in the requested field. A matching record with ordinary text instead of a measurement does not become a zero-valued observation.
Set up the measurement log
Enter this original practice dataset in A1:D11. Type pending as ordinary text. Physically leave empty the two cells labeled leave empty below; do not type that label or use an empty-string formula.
| Run | Batch | Shift | Minutes |
|---|---|---|---|
| 101 | 10 | 1 | 4 |
| 102 | 10 | 1 | 8 |
| 103 | 10 | 2 | 12 |
| 104 | 10 | 2 | pending |
| 105 | 10 | 2 | leave empty |
| 106 | 20 | 1 | 6 |
| 107 | 20 | 1 | 6 |
| 108 | 20 | 2 | 6 |
| 109 | 30 | 1 | pending |
| 110 | 30 | 2 | leave empty |
Put Batch in F1 and numeric 10 in F2. Use F1:F2 as the criteria range. Place output formulas away from both data and criteria, such as H1 downward. Examples use English names and semicolon separators.
Example 1: calculate spread in minutes
=DSTDEV(A1:D11;"Minutes";F1:F2)
Expected result: 4. The three numeric observations are 4, 8 and 12. The mean is 8, and squared deviations total 16 + 0 + 16 = 32. Dividing by 3 - 1 gives sample variance 16; taking its square root gives standard deviation 4.
The five matching records do not make this a five-value numeric sample. Check both counts beside the result:
| Formula | Expected result |
|---|---|
=DCOUNT(A1:D11;;F1:F2) |
5 records |
=DCOUNT(A1:D11;"Minutes";F1:F2) |
3 numeric measurements |
Report 4 minutes, not 4 square minutes or a range containing every possible run. The mean and standard deviation summarize the selected values; they do not impose bounds on future observations.
Example 2: narrow the selection to a shift
Put Batch and Shift in F4:G4, with numeric 10 and 1 in F5:G5. Conditions on that one criteria row apply together.
=DSTDEV(A1:D11;"Minutes";F4:G5)
Expected result: sqrt(8), approximately 2.828427 minutes. The selected values are 4 and 8. They have mean 6, squared deviations totaling 8, and a sample denominator of 2 - 1.
Do not calculate this by subtracting the removed value from the original standard deviation. The selected mean and denominator have both changed. For more complex filters, see DSUM’s AND/OR criteria examples.
Example 3: convert the measurement unit consistently
In a copy of the sheet, convert the three numeric Batch 10 times in D2:D4 to seconds: replace 4, 8 and 12 with 240, 480 and 720. Rename D1 to Seconds, then use:
=DSTDEV(A1:D11;"Seconds";F1:F2)
Expected result: 240 seconds, which is 60 times the original 4-minute standard deviation. The converted sample variance is 57,600 square seconds, and its square root is 240.
This temporary sheet deliberately converts only Batch 10. Do not widen the criteria to other batches until every numeric time has been converted to the same unit. Keep the original practice sheet unchanged for the remaining checks.
Troubleshooting zero and too few measurements
The official reference gives these documented boundaries, which have not been executed in Collabora here:
- No matching records: F2 set to 99 leads to a documented 0.
- Two matching records with no numeric Minutes values: F2 set to 30 also leads to a documented 0.
- One numeric observation: with Run in F7 and 101 in F8,
=DSTDEV(A1:D11;"Minutes";F7:F8)has documented outcome #NUM!. The source also specifies #NUM! when several matching records contain only one numeric value in the measured field.
Batch 20 provides a different zero: its measured values are 6, 6 and 6, whose independently calculated standard deviation is zero. Use DCOUNT to separate absent measurements from an actual constant sample.
The reference’s one-record and no-numeric statements overlap if the sole matching record is nonnumeric. This guide does not assign an outcome to that combined case. Formula-empty strings, numeric-text coercion and mixed input errors are outside the examples.
Related functions and interpretation
DVAR reports the same sample’s variance: 16 square minutes. DSTDEVP describes population standard deviation and gives approximately 3.265986 minutes for the same three values. Neither choice fills missing measurements; decide whether the observations represent a sample or the entire population you intend to describe.
References and verification
- Collabora Help: DSTDEV contract, syntax and boundaries.
- Shared database arguments and criteria.
- Editorial testing standards.
Selections, counts, squared deviations and displayed standard deviations were checked independently with rational arithmetic and Python statistics. The documented zero/error outcomes are not independent statistical computations or native test results. No native execution or screenshot is claimed.