_n not referring to observation number

[Taewan Roh]

Hello!

I have run into an issue where _n does not refer to the observation number.

The commands and results I have seen are listed below:

. gen id = _n

. format id %22.0f

. list id in 16777213 / 16777219


---ob no----+-----id-------+
16777213. | 16777213 |
16777214. | 16777214 |
16777215. | 16777215 |
16777216. | 16777216 |
16777217. | 16777216 |
16777218. | 16777218 |
16777219. | 16777220 |
---------------+---------------+

Is there a particular reason why this would be happening? I have tried other methods too such as the one below but always get similar results.

. gen id = 1

. gen sumid = sum(id)

. format id sumid %22.0f

. list id sumid in 16777213 / 16777219

---ob no----+-id---sumid---+
16777213. | 1 16777213 |
16777214. | 1 16777214 |
16777215. | 1 16777215 |
16777216. | 1 16777216 |
16777217. | 1 16777216 |
16777218. | 1 16777218 |
16777219. | 1 16777220 |
----------------+-----------------+

My ultimate goal would be to generate a variable with a unique id for each of the observations. Would there be any way to work around this?

Thank you for your time and help!

[eric_a_booth]

It's a precision issue.
See this posting for some background (http://blog.stata.com/tag/precision/) and -help precision-, but consider this example where specifying 'double' format for the variable "id" works:

********************
clear
set obs `=10^8'

**doesnt work:
. gen id = _n
. format id %22.0f
. list id in 16777213 / 16777219


**works:
g double id2 = _n
. format id2 %22.0f
. list id2 in 16777213 / 16777219
********************

[Taewan Roh]

Thank you so much! It works perfectly now.

I was looking in all the wrong places...