LOGINV in WPS Spreadsheet: Lognormal Quantiles on the Original Scale

=LOGINV(probability, mean, standard_dev)

Mean and standard_dev are the mean and standard deviation of ln(x). All original controls use a positive log-scale standard deviation and probabilities strictly between zero and one.

The WPS formal error remarks exclude probabilities at or outside 0 and 1, while the walkthrough describes an inclusive [0,1] interval. This guide retains the stricter interior domain for examples and makes no observed endpoint-error claim. The phrase “new WPS version” in the lesson does not identify a numeric build or establish the function’s introduction date.

Enter 0.5, 1 and 0.5 in A2:C2, then use =LOGINV(A2,B2,C2) in D2.

The standard-normal median is zero. The independent transformation is exp(mean + standard_dev*z), so the expected result is exp(1), approximately 2.71828182845905. That is the original-scale median; the argument mean=1 remains a log-scale mean.

For p=0.75 with the same parameters, the standard-normal quantile is approximately 0.674489750196082. Exponentiating 1+0.5*0.674489750196082 gives approximately 3.80853604483271.

These calculations use an independent normal inverse followed by exponentiation. Rounded values should not be used as claims about WPS’s stored precision.

Take a candidate original-scale quantile x, form z=(ln(x)−mean)/standard_dev, and independently evaluate the normal CDF at z. It should reproduce the requested p within the chosen numerical tolerance. This checks the defining inverse relationship rather than comparing only printed final digits.

For the 25th and 75th percentiles above, the logarithms are equally far below and above 1. Their product is exp(2), approximately 7.38905609893, and their geometric midpoint is exp(1). Their arithmetic midpoint is not the median. This is a useful check when interpreting a positively skewed distribution on its original scale.

A theoretical quantile is not an empirical percentile of a finite dataset or a confidence interval for an estimated parameter. The model and its parameters need their own evidence.

The official walkthrough uses p=0.022, log-mean 4.1 and log-standard-deviation 1.8, displaying 1.607429852. The function dialog explicitly identifies the parameters on the ln(x) scale, supporting the transformation despite loose variance wording elsewhere.

The lesson documents nonnumeric-input and nonpositive-standard-deviation errors. The original controls avoid malformed inputs and all extreme probabilities; they do not establish endpoint behavior, implementation convergence or modern alias availability. Check that a 75% probability is stored as 0.75, not 75, and that the result is being interpreted on the original scale.

For a normal-model inverse without exponentiation see NORMINV. PERCENTILE instead returns a threshold from observed data.

The official WPS lesson is dated 2019-05-16 and explicitly labels WPS Office for Windows Personal. Its numeric build is unspecified. This documents the function by that date; it does not establish first introduction, the last functioning build, other-platform parity or a retirement/deprecation history.

The official lesson was rechecked on October 19, 2026. Original examples are documentation-derived independent calculations, not WPS or other spreadsheet-engine executions. Numeric reference values do not certify application precision or statistical model suitability.