Multiplication errors using gen

[Chris Martin]

I'm using the code below to generate a new variable that is the unique id multiplies by 100. However, the result is not the correct value.

gen entryid=.
format entryid %12.0g
replace entryid=(ExternalReference * 100)

For instance ExternalReference 1000863 does not end up as 100086300, but rather 100086304

Does anyone know what could cause this?

Chris

[FernandoRios]

Hi Chris,
The answer to your problem resumes to One word: Precision.
By default, Stata creates all new variables in float format, which stores numbers using 4bytes of information.
see help data_types
help precision
If the number is too long, it will round it up to the nearest binary approximation. Try using "double" instead.

HTH
Fernando

[Bruce Weaver]

I was going to suggest using "long" rather than "double". For the example Chris gave, either will work.

clear *
input ExternalReference
1000863
end

generate long id1 = .
generate double id2 = .
format id1 id2 %12.0g
replace id1 = ExternalReference * 100
replace id2 = ExternalReference * 100
list

Output:

. list

     +----------------------------------+
     | Extern~e         id1         id2 |
     |----------------------------------|
  1. |  1000863   100086300   100086300 |
     +----------------------------------+

[Chris Martin]

Thank you. This solved the problem. It would be nice if Stata outputted a warning message to tell the user that it rounded up numbers.
Chris