CHISQ.INV.RT and CHIINV in ONLYOFFICE Spreadsheet Editor

• Introduced: Original introduction release is not established by the sources checked.
• Available: Documented in current unversioned ONLYOFFICE function help, checked October 18, 2026.
• Implementation inspected: Official SDKJS v10.0.0.169, commit 3679e0375a758d7dc59d31337a45461be3525ed8. This establishes a source snapshot, not identical behavior in every installed editor.
• Retired or deprecated: Retirement or formal deprecation is not established by these sources. A legacy spelling is not automatically a removed function.
• Testing: Examples are independently calculated or labeled source predictions. No ONLYOFFICE application or calculation-engine execution was performed. No native screenshot or general import/export guarantee is claimed.

Use English names and comma separators as shown; adapt separators outside quoted text if your locale requires semicolons. Keep full precision in calculations and round only the displayed summary.

=CHISQ.INV.RT(probability, deg_freedom)

Exactly two arguments are required. This is an inverse distribution function: the input is a probability and the output is a nonnegative statistic threshold. Do not enter a computed chi-square statistic in the probability position. A probability of 5% is 0.05, not 5.

The legacy CHIINV name also calculates a right-tail inverse. It belongs with CHISQ.INV.RT, not with CHISQ.INV’s left-tail semantics.

1. On a fresh worksheet put 0.05 in A2 and 2 in B2.
2. Enter =CHISQ.INV.RT(A2,B2) in C2.
3. The independent expectation is 5.991464547108.
4. Check the direction by evaluating =CHISQ.DIST.RT(C2,B2); its mathematical target is A2, or 0.05.

For two degrees of freedom, the right-tail probability is exp(−x/2). Solving exp(−x/2)=0.05 gives x=−2 ln(0.05). That directly matches the supplied right-tail probability. This logarithmic result is an independent calculation; it is not a recorded ONLYOFFICE solver result.

A threshold is a location on the reference distribution. It is not a probability, and formatting it as a percentage would obscure its meaning. Label the output “chi-square threshold” and keep the probability and df in separate cells.

At probability 0.5 and df=2, either tail’s inverse gives the median 1.386294361120, since −2 ln(0.5)=1.386294361120. Equality at the median does not mean left and right inverses are interchangeable elsewhere.

At right-tail probability 0.95, the threshold is only 0.102586588775. A large right-tail area leaves a small statistic below its boundary. By contrast, right-tail probability 0.05 gives the larger 5.991464547108 threshold.

For probabilities strictly between 0 and 1, the mathematical identity is:

CHISQ.INV(p, df) = CHISQ.INV.RT(1−p, df)

This is an equation about the distribution, not a promise of identical floating-point results for extreme probabilities. Keep the original probability at full precision. If a tiny probability is rounded to zero, or a probability near 1 is rounded to exactly 1, the inverse problem changes to a boundary case or an invalid input.

Official help describes probabilities strictly between 0 and 1 and df from 1 through 10^10. The inspected right-tail code actually permits probability 1 and rejects probability 0; mathematically, the accepted endpoint corresponds to statistic 0. Its wrapper rejects df below 1 but does not include the explicit 10^10 upper guard found in CHISQ.INV. Stay within the documented df range rather than interpreting the missing guard as supported unlimited degrees of freedom.

These endpoint statements describe the source guard and mathematical boundary, not a native application observation. The comparison table uses probability 0 as a source-predicted numeric error, conventionally #NUM!. Use interior probabilities for routine threshold calculations.

Degrees of freedom are parsed as an integer. Enter the intended whole value, rather than relying on truncation of a fractional input. A valid numeric df does not establish that it is the correct df for your test; for example, fitted parameters or a different table shape can change it.

The inspected inverse helper first attempts to bracket the target with up to 1000 expansions, then uses interpolation with a bracket-based fallback for up to 500 refinement iterations. A failure to bracket is reported through an error flag. Reaching the refinement limit is not, by itself, converted into that same failure flag in the helper.

A source comment about an iteration count is therefore not a guarantee that every successful return achieved a requested accuracy. Verify important results by substituting the returned threshold into the matching distribution function and comparing with the requested probability. Use an appropriate tolerance and do not require exact binary equality. Extreme tails, huge df and imported text/array coercions were not tested in the application for this guide.

These are mathematical expectations or identified implementation predictions, not observed application results.

The legacy name CHIINV is covered here with the modern name. CHIINV and CHISQ.INV.RT use the same right-tail objective, inverse helper and shown validation logic in separate definitions. This is calculation-path evidence, not proof that metadata, serialization or every edge case is identical. Do not change a left-tail function to a legacy right-tail name merely because both contain “CHI.”

Modern dotted names are marked as newer-format functions in the inspected SDKJS metadata. That flag does not establish their introduction date or guarantee support in older file readers. Test a small known-value example when moving a workbook between applications.

Related guides: EXP and LN reproduce the two-degree-of-freedom checks; SUM combines statistic contributions; COUNT helps inspect numeric inputs.

• Official CHISQ.INV.RT help, checked October 18, 2026.
• Pinned SDKJS implementation, including chi-square helpers and argument validation.
• Official CHIINV help, for the related legacy spelling.