ONLYOFFICE AVEDEV: Average Absolute Deviation with Worked Examples

• Introduced: The first introduction release and date are not established by the checked sources.
• Availability: Currently documented in ONLYOFFICE’s rolling function help; no final supported-version endpoint is established.
• Scope: Shared Spreadsheet Editor documentation. Exact availability in individual Docs, Desktop and mobile builds is not established by this page.
• Checked: October 14, 2026, against the official AVEDEV reference. Examples were independently calculated and compared with the vendor screenshot; these cases were not tested in the native application.

=AVEDEV(number1, [number2], ...)

The official reference documents one required numeric argument and optional further arguments, up to 255. Its example combines a cell range with separately supplied numbers. The brackets describe optional arguments; do not type them into a formula.

AVEDEV first finds the arithmetic mean, then takes the absolute distance of every included number from that mean, and finally averages those distances. It is not the average of the original numbers and is not a standard deviation.

For the input 8, 10, 10, 12, the mean is 10. The absolute distances are 2, 0, 0, 2, whose average is 1. If the original measurements are in millimeters, this deviation is also in millimeters.

On a blank worksheet, put Measurement, Signed deviation and Absolute deviation in A1:C1. Enter 8, 10, 10, 12 in A2:A5. In E1 enter Mean, in E2 enter =AVERAGE(A2:A5), and in E4 enter =AVEDEV(A2:A5).

Expected E2 is 10, and expected E4 is 1. Now show the intermediate steps:

• B2: =A2-$E$2
• C2: =ABS(B2)

Fill B2:C2 through row 5.

In E5 enter =AVERAGE(C2:C5); expect 1, matching E4. In E6 enter =AVERAGE(B2:B5); expect 0. That zero is why averaging the signed deviations would fail to measure this spread.

AVERAGE supplies the center, and ABS removes the sign from each distance. Keep the dollar signs in $E$2 so filling down does not move the mean reference.

Use the empty G:J area. Put Measurement, Signed deviation, Absolute deviation and Mean in G1:J1. Enter 8, 10, 10, 20 in G2:G5 and =AVERAGE(G2:G5) in J2. The expected mean is 12.

Enter =G2-$J$2 in H2 and =ABS(H2) in I2; fill H2:I2 through row 5. Expected signed deviations are -4, -2, -2, 8, and their absolute values are 4, 2, 2, 8. Put =AVEDEV(G2:G5) in J4 and =AVERAGE(I2:I5) in J5; both independently expected results are 4.

The changed observation moves the mean as well as increasing the distances. Do not continue subtracting the first batch’s mean of 10 after changing the data.

This measure uses the mean as its center. It is different from a median-centered absolute-deviation measure, so specify the actual statistic when comparing reports from other tools.

Keep the two batches intact. In A8:A11 enter 108, 110, 110, 112, and put =AVEDEV(A8:A11) in E8. Expect 1: adding 100 to every observation changes the mean but not its distances.

In G8:G11 enter 16, 20, 20, 24, and put =AVEDEV(G8:G11) in J8. Expect 2: doubling every measurement doubles each absolute distance. This is a useful unit-conversion check, not a claim about every floating-point boundary.

ONLYOFFICE’s displayed example uses 78, 98, 123, 45, 70, 67, 3, 9, 289. On a separate sheet, enter those nine numbers in A1:A9 and use =AVEDEV(A1:A9) in C1. The independently calculated result is 1496/27, approximately 55.4074074074, agreeing with the screenshot’s rounded 55.40740741.

The short help page does not specify a full blank, Boolean, text-in-range or error-propagation matrix. The examples here deliberately use complete numeric data. A blank is not automatically a valid zero observation; check the source before replacing missing measurements.

The page also does not enumerate exact error tokens for an empty effective dataset or extreme values. Keep headers out of the selected numeric range and inspect units before interpreting the result. DEVSQ instead sums squared deviations, which gives squared units and greater emphasis to large distances.