[EpiData-list] Bug in EpiData Analysis

[epidata-list at lists.umanitoba.ca]

This is an email to the folk at EpiData.


In my examples below I'm using the data file bromar.rec, one of the example 
data files that come with EpiData.

The variable km has some missing data (.).  This data seems to be correctly 
treated as missing values for some commands but not for all.  One example 
is the StatTables command.  Below is an example of the command I typed into 
EpiData and the resulting output:
. StatTables sex /idr=km
SEX             N       NKM     P10KM   P90KM
Female  490     423     42.00   .
Male            3537    3197    32.00   .

As you can see the missing values are included in the calculation for the 
90th percentile. It's obviously a bug. However you may not detect the bug 
if you're interested in the interquartile range:
. StatTables sex /iqr=km
SEX             N       NKM     P25KM   P75KM
Female  490     423     67.00   265.00
Male            3537    3197    42.00   265.00
If you do a "describe" you can see that 265 is definitely not the 75th 
percentile.


Suzanna Vidmar
Senior Research Officer

Clinical Epidemiology and Biostatistics Unit
Royal Children's Hospital
Flemington Rd
Parkville Victoria 3052

telephone: +61 3 9345 6372
facsimile: +61 3 9345 6000