PERCENTILE.INC in Collabora Office: Inclusive Endpoints and Interpolated Threshold Audits

• Introduced: original Collabora release and launch date are not established by the reviewed sources.
• Availability: currently documented in Collabora Office 26.04 desktop help.
• Platform scope: online and mobile deployments are not inferred from this desktop-help release.
• Checked: October 14, 2026. Function-specific documentation and independent calculations; no native Collabora execution.

=PERCENTILE.INC(data;alpha)

Both arguments are required. Data is a range or array; alpha is the requested fraction. The dedicated help distinguishes the inclusive alpha range from PERCENTILE.EXC’s exclusive one.

These examples use complete numeric data. The PERCENTILE.INC section does not repeat the range-text exclusion rule found on plain PERCENTILE, so mixed inputs are not silently assigned that other spelling’s rules.

The independent numeric model uses position 1+(n-1)×alpha in sorted observations. Fractional positions are checked by linear interpolation. This equation makes the examples reproducible without claiming native execution or a complete precision specification from the compact reference.

Enter 2,8,14,26,50 in A2:A6:

The minimum, median and maximum controls directly match the source’s named points. The mean is 20, which is not the median 14.

These values describe the same dataset at different positions. They are not five separate groups, and the function does not assign each row to a percentile bucket.

=PERCENTILE.INC(A2:A6;0.9)

Independent expected result: 40.4. The position is 4.6, between 26 and 50, so the calculation is 26+0.6×24.

By comparison, the 10th-percentile model has position 1.4 and gives 4.4, from 2+0.4×6. The numerical gaps are unequal, so different percentile intervals need not span equal measurement differences.

Duplicate every original observation, producing 2,2,8,8,14,14,26,26,50,50. The mean and median remain 20 and 14, but the 25th-percentile model now uses position 3.25, still within the two repeated 8s, and gives 8.

The 10th-percentile model changes more noticeably: position 1.9 lies between two 2s, giving 2 instead of 4.4. A repeated distribution can preserve some summaries while changing finite-sample interpolation positions.

This is an arithmetic counterexample, not permission to duplicate records. Record counts and observation provenance still matter.

In C2:C7 enter 2,8,14,26,26,50. The upper-quartile model position is 4.75, between equal values 26 and 26, so the cutoff remains 26.

A rule accepting values at least 26 includes three of six observations. Strictly greater than 26 includes only one. State the comparison operator as well as the percentile when a report selects records.

A percentile cutoff does not define a unique fixed-size selection when ties exist. RANK.EQ provides another example of tied-value selection policies.

Add 10 to every original observation. The transformed list is 12,18,24,36,60, and the independent 90th-percentile cutoff becomes 50.4, ten above the original.

Multiplying the original list by three instead produces cutoff 121.2. Keep units in the report; a percentile value is expressed in the measurement units, whereas a percentage rank is a dimensionless position.

The source identifies COM.MICROSOFT.PERCENTILE.INC outside the cited OpenFormula 1.3 standard. This does not imply unavailability in Collabora or universal file-interchange success.

PERCENTILE.EXC uses a different positional convention and a sample-dependent interpolation window. Invalid-alpha errors, Boolean/text conversion, empty or singleton data and extreme numerical precision remain unverified here. PERCENTILE retains its own current documentation rather than an inferred retirement notice.

• Collabora Office 26.04: PERCENTILE.INC.
• Editorial testing standards.

Complete function-specific documentation reviewed. Worked outputs are independent mathematical models or explicitly documented rules, not application screenshots or native test results. Precision, type and error boundaries retain their stated limits.