Contact

RANDSTDIST Function (Gnumeric)

Statistical Intermediate Gnumeric

Generate skew-t draws with Gnumeric RANDSTDIST, supplying both required arguments and distinguishing shape from output moments.

Verification status: See the evidence and test scope described in this guide. How formula examples are checked

RANDSTDIST generates a skew-t random variate. Its shape parameter controls the distribution’s asymmetry; it is not simply the output’s skewness, mean, or standard deviation. The symmetric case is obtained with an explicit shape value of zero.

Syntax and a version-specific help mismatch

RANDSTDIST(df, shape)

Use positive degrees of freedom df and a finite real shape. The 1.12.57 help describes shape as optional with default zero, and the callback contains a fallback value. However, the same version’s function descriptor registers two required numeric arguments. Supply both arguments rather than relying on the help’s advertised omission.

A shape of zero selects the ordinary t generator. A nonzero shape uses a skew-normal draw scaled by the square root of df divided by an independent chi-square draw. Positive and negative shape values alter the asymmetry; they do not force every draw to have the same sign.

Worked examples

=RANDSTDIST(5,0) requests the symmetric t case. =RANDSTDIST(5,2) and =RANDSTDIST(5,-2) request skew-t draws with opposite signs of the shape parameter. Each returns a run-specific random value; re-entering the formula does not reproduce a particular decimal result.

Use separate columns labeled “df,” “shape,” and “draw” when making a small experiment. Keep df=5 fixed and compare shape=0, 2, and -2 over many draws. One positive or negative observation is not evidence that the generator has the wrong skew direction. Assessing distributional shape requires a suitable sample and statistical analysis beyond these dispatch checks.

=RANDSTDIST("bad",2) supplies text where a numeric argument is required and produces #VALUE!. The native =RANDSTDIST(5) check returned #N/A, confirming that omission is not accepted in this version despite the help text. The explicit-zero follow-up =RANDSTDIST(5,0) returned about -1.4813980303001273 in one run. The two skewed examples returned about 0.6792250557516948 and -0.8070665672429073; these observations are not fixed expected values.

Advanced use: location and scale

If a random variate is stored in A2, =10+2*A2 applies a location shift of 10 and a positive scale factor of 2. It does not generally produce a distribution with mean 10 and standard deviation 2. In particular, skewness can shift the mean, and t-family moments depend on the degrees of freedom.

For the symmetric case with df=5 and shape=0, the theoretical mean is zero and variance is 5/3. Those are distribution properties, not outputs of one RANDSTDIST call. Do not infer a finite variance for every positive df. When a model requires particular moments, derive and check the appropriate parameter transformation rather than treating the argument names as moment controls.

Errors and troubleshooting

The callback passes degrees of freedom to the underlying generators rather than supplying a complete worksheet-level domain check. Keep df strictly positive; this guide does not execute or promise a specific error for invalid df. Avoid extreme parameters without additional numerical testing.

The function is volatile. Reference a stored draw if several model outputs must share the same scenario, and preserve a values-only snapshot when comparing runs. No worksheet seed argument is provided. A small set of successful native draws validates neither statistical independence nor tail accuracy.

Version and verification

Official release notes introduce RANDSTDIST in Gnumeric 1.9.15. The guide’s contract and arity discrepancy are pinned to official 1.12.57 source. The selected examples were recalculated by the actual Gnumeric 1.12.57 engine under C.UTF-8 on 17 October 2026. No invalid-df execution, GUI screenshot, newer-version equivalence, retirement, or last-supported-release claim is made.

Official references