On 7/19/2005 9:28 AM, Clark Allan wrote:> for the stats gurus
>
> does anyone know if there exists a general formula relating the median
> of a continuous distribution to its moments. the distribution could be
> skewed or symmetric and is definitely not normal.
Not in general, and probably not any more practically useful than
F^(-1)(0.5).
> the reason for asking is since the median of the particular distribution
> that i am interested in is difficult (probably impossible) to obtain.
> the median depends depends on an incomplete gamma distribution. the
> moments however can be obtained fairly easily.
R can calculate the incomplete gamma function (see ?pgamma), so that's
not necessarily a stumbling block.
Duncan Murdoch