CORREL and PEARSON in Calligra Sheets: Correlation Examples

CORREL measures the strength and direction of a linear relationship between two corresponding numeric data series. It returns the Pearson correlation coefficient r, theoretically between −1 (perfect decreasing line) and +1 (perfect increasing line). A value near 0 describes a weak linear association, not proof of no relationship. Correlation is not causation.

=CORREL(A2:A6;B2:B6)

For the concrete worksheet below the independently calculated expected r is 0.904194430179. KDE also registers PEARSON with the same implementation, so =PEARSON(A2:A6;B2:B6) has the same mathematical expectation. PEARSON is a registered equivalent name covered in this guide. This is a source-backed equivalence, not a claim of native execution.

=CORREL(range1;range2)
=PEARSON(range1;range2)

• range1 (required): first numeric cell range/array (KDE XML specifies numeric Double and range=true).
• range2 (required): second numeric cell range/array of observations corresponding positionally to range1.
• Number of arguments: exactly two according to pinned C++ setParamCount(2), with array inputs accepted by setAcceptArray().
• Defaults or optional arguments: none in the inspected registration/XML.
• Order: swapping the two ranges leaves r unchanged mathematically. Preserve pairing regardless of order.
• Result type: numeric floating-point coefficient when mathematically defined. Both input series must have nonzero variation for correlation to be defined.

KDE explicitly documents CORREL(range1; range2) and PEARSON(range1; range2); KDE’s pinned XML function definitions identify both as equal-purpose cell-range functions. In the 26.08.2 C++ registrations, CORREL maps to func_correl_pop; PEARSON maps to that same func_correl_pop. The implementation divides covariance by the product of two population-standard-deviation values, and explicitly tests for zero standard deviations.

On a new sheet, enter these numbers in A1:G6. Row 1 is text labels; rows 2–6 are numeric. Keep the results outside this input area (for example in I2 and below). The helper columns deliberately transform the same pairs so you can check whether a formula responds as expected.

Each row is one paired observation: A2 corresponds to B2, A3 to B3, and so on. Never sort one measurement column without sorting its paired column. Use a period as the decimal separator only in a locale configured for that convention. The formulas below use the semicolon (;) separator shown in KDE’s English-language function documentation; localized applications may use a different display/entry convention.

To reproduce a row in any example table, click an unused result cell, type the formula beginning with =, press Enter, and compare its display to the independently calculated expected value. Increase displayed decimal places when checking precision. The numerical values in this guide were not obtained from running Calligra Sheets.

These are mathematical expectations for the displayed inputs, not results observed in native Calligra. Values after ≈ are rounded for display.

Manual arithmetic cross-check: x̄ = 3; ȳ = 4.2; the sum of paired products of deviations is 11; the sums of squared deviations are 10 and 14.8. Therefore r = 11 / SQRT(10 × 14.8) = 0.904194430179. This independent formula agrees with the C++ covariance/standard deviation relationship; it is not a screenshot or application test.

Suppose column A records advertising spend units and column B records resulting sales units, with each row the same observation period. Put the slope in I2, the intercept in I3, and the correlation in I4:

I2: =SLOPE(B2:B6;A2:A6)
I3: =INTERCEPT(B2:B6;A2:A6)
I4: =CORREL(A2:A6;B2:B6)
I5: =I4^2

The independently calculated expected results are 1.1, 0.9, 0.904194430179 and 0.817567567568 respectively. In a single-predictor ordinary-least-squares regression with an intercept, r² describes the fraction of sample y variation explained by the fitted line. That does not establish a causal effect, a reliable future forecast, or a strong nonlinear fit.

For an explicit check without built-in correlation use an unused cell:

=11/SQRT(10*14.8)

The expected output is approximately 0.904194430179. These values may be rounded slightly by displayed cell format.

• Constant input: For =CORREL(A2:A6;G2:G6) the second range is always 5, so mathematical r is undefined. The pinned C++ wrapper explicitly returns an internal division-by-zero error when a population standard deviation is zero; the visible GUI error string was not observed.
• No corresponding numeric pairs or mismatched counts: func_correl_pop calls func_covar, which checks argument numeric counts and can emit an internal value error. The exact displayed error and pair filtering must be verified in native Calligra; do not copy Excel’s error behavior verbatim.
• Very few observations: r requires nonzero variation in both series; two distinct nonconstant points can define ±1 mathematically, but real-world evidence needs more observations.
• Outliers and curved relationships: Pearson r is sensitive to extreme observations and need not describe a strong nonlinear relationship. Inspect a scatter plot before interpreting it.
• Unexpected sign: verify that values are paired in the same rows and neither series was sorted independently.
• Very large or almost constant values: floating-point cancellation can affect computed correlation; native precision/overflow boundaries were not tested.

Shared troubleshooting checklist

1. Confirm the entered formula begins with = and the function name is spelled exactly as shown.
2. Follow the KDE semicolon syntax when using an English-style installation (; rather than Excel’s common ,). See KDE’s formula-entry guidance; adjust for your application’s locale where necessary.
3. Ensure both ranges represent the same observations in matching row order; the example uses five numeric points, not five random numbers from separate contexts.
4. Check for numbers stored as text, blank cells, Boolean values and nonnumeric entries. The handbook/source do not establish all native pair-exclusion and coercion cases tested here; normalize source data or validate them in your installed build.
5. Check for constant or nearly constant series. Correlation and regression denominators can be zero or poorly conditioned.
6. If the result is visibly rounded, increase displayed decimal places before judging it incorrect. Floating-point arithmetic can produce tiny rounding differences.
7. For a parsing failure, inspect argument separators and region-specific decimal characters before changing formula logic.
8. For an unexpected error from a range, try the five-number all-numeric reproduction above, then restore your actual data one cell at a time. Do not assume a particular error label without a native test.

Microsoft’s Excel CORREL reference independently documents the same statistical purpose using CORREL(array1, array2) with an English-Excel comma. KDE documents CORREL(range1;range2) with semicolons for its shown locale. KDE’s PEARSON name is also implemented by the same Calligra function. This is name/mathematics-level compatibility only: we have not imported an XLSX file into Calligra or verified handling of text, blanks, Boolean values, matched filtering, locale translation, or the exact error displayed on either platform. Microsoft’s published input-handling rules are Excel evidence, not proof of Calligra behavior.

• Official KDE CORREL documentation, PEARSON documentation, COVAR documentation, STDEVP documentation, SLOPE documentation and INTERCEPT documentation.
• KDE C++ statistical implementation, pinned 26.08.2 source and KDE statistical XML definitions.
• Microsoft Excel CORREL reference — comparison only; Calligra formula entry conventions.

Research snapshot: KDE Gear / Calligra 26.08.2 (released October 8, 2026); checked October 13, 2026. The KDE application release listing identifies 26.08.2 as the current release on the checked date, and KDE’s release announcement records the release date. The handbook is a useful syntax reference but is not itself proof that every old handbook example matches every modern binary. This guide also compares the exact 72123cc643f6625351139e7df294544255a1056f source revision.

Platform: The cited upstream release and package information concerns Calligra primarily distributed for Linux, including distribution packages and Flathub. Which function is available in your installed binary depends on its package, version and enabled modules. We did not test Windows, macOS, mobile, or individual distribution builds, so this page does not certify identical results on those platforms.

Validation labels: OFFICIAL_DOCUMENTATION_CHECKED = described by KDE’s documentation/XML; SOURCE_LOGIC_INSPECTED = the named functions and argument counts were found in the pinned C++ source; INDEPENDENT_ARITHMETIC = output computed using ordinary mathematics outside Calligra; NATIVE_CALLIGRA_EXECUTION_NOT_RUN = no actual Calligra spreadsheet, screenshot, error dialog, or imported Excel workbook was tested. Treat boundary cases specifically marked unverified as test cases, not promised visible outputs.