[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: RE: memory error 910 using Fisher's exact test

From   "Nick Cox" <>
To   <>
Subject   st: RE: memory error 910 using Fisher's exact test
Date   Sun, 14 Oct 2007 22:10:21 +0100

It sounds as if at least one variable is ordinal (as
it is a quintile (class)), and perhaps both. 

In this situation Fisher's exact test would not 
be matching what you know about your data in any case. 

Also if you have -hourwkquintile- you may have  
the underlying data. 

More positive advice would seem to require more detail
on your data and research question. 


Julie Darnell
> I am using Stata 9 and set my memory to 950m (the maximum 
> allowed on my 
> computer).
> I attempted to execute the syntax below:
>       tab hourwkquintile q25yearoffoundingnew, exact row col
> where there are 5 groups each for hourwkquintile variable and the 
> q25yearoffounding variable, for a 5X5 table of 751 total cases.
> The command terminated with error 910, exceeded memory limits.
> In these cases, the Stata help for tabulate suggests using 
> exact(2). I 
> therefore re-ran the command:
>       tab hourwkquintile q25yearoffoundingnew, exact(2) row col
> but the addition of (2) (or any larger number up to 25 after 
> exact) does 
> not allow the command to execute fully to its completion. It 
> gets stuck 
> during the stage 3 enumeration.
> Collapsing the categories to avoid having expected cell 
> counts of 5 or 
> fewer cases is not desired on conceptual grounds.

*   For searches and help try:

© Copyright 1996–2015 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index