trying to use rangejoin - never had an issue before

[Rose Matthews]

Hi I've used rangejoin before and never had issues.
Somehow this time it won't work


Dataset1 : admissiondate uniqueid gender - variables
Dataset2: surgerydate uniqueid gender

UniqueId in both datasets are the same. However in dataset2, there are several duplicate uniqueids and therefore not unique values. I want to try and eliminate those duplicate uniqueids that have incorrect information.

Aim: To bring admission date from Dataset1 onto Dataset2 for each uniqueid that is found in both Dataset 1 and Dataset2 using UniqueID and Gender otherwise it is tagged as missing.
Following this I will then check which ones have admissiondate=surgerydate. If != that it is tagged and then dropped.

clear all
use "dataset1"
rangejoin admissiondate using "Dataset2", by (uniqueid gender)

Error: was expecting a numeric variable, a number or a system missing value for the interval low

I used the above same code for another rangejoin and never had issues

admissiondate - stata date
uniqueid - double
gender double

[Clyde Schechter]

I used the above same code for another rangejoin and never had issues

I really doubt that. It's clearly not compatible with -rangejoin-'s syntax. You must specify a lower and upper bound for the rangejoin following admissiondate and before using.

Also, the variable that keys the range must be a variable in the using data, not the master. So you can't use admissiondate there.

rangejoin surgerydate . . using "Dataset2", by(uniqueid gender)

But, if you don't want to constrain which surgerydate values get paired up with which admissiondate values, then why use -rangejoin- at all. Just do

joinby uniqueid gender using "Dataset2"

I suspect, though, that you really will want to constrain which surgery dates get paired with which admission dates. Say, maybe, within 90 days of each other? That would be

gen lb = admissiondate - 90
gen ub = admissiondate + 90
rangejoin surgerydate lb ub using "Dataset2", by(uniqueid gender)

Always remember: the key variable for the rangejoin comes from the using data set, and the bounds come from the master data set.