Creating a %15.0f variable

[Paula de Souza Leao Spinola]

My dataset structure is as follow:

uf    mun    dist    subdist    set
11    15    5    0    1
11    15    5    0    1
11    15    5    0    1
11    15    5    0    1
11    15    5    0    1
11    15    5    0    1
11    15    5    0    1
11    15    5    0    1
11    15    5    0    1
11    15    5    0    1

My row identifier should be a 15 digit one which I should get through the code below:

gen double setor = 100000000000000*uf + 1000000000000*mun + 10000000*dist + 100000*subdist + 1000*set
format setor %15.0f

However, Stata gives me back the variable below, which is not displayed as %15.0f. How could I get it right?

setor
1.11500005e+15
1.11500005e+15
1.11500005e+15
1.11500005e+15
1.11500005e+15
1.11500005e+15
1.11500005e+15
1.11500005e+15
1.11500005e+15
1.11500005e+15

[Nick Cox]

You're miscounting somewhere. The created values have 16 digits, so Stata is ignoring the specified format. You just need another. Note that a format too large is not usually a problem, but a format too small often is.

As your example is just repeating the same calculation 10 times, I slimmed it down.

. clear 

. input uf    mun    dist    subdist    set

            uf        mun       dist    subdist        set
  1. 11    15    5    0    1
 2. end 

. gen double setor = 100000000000000*uf + 1000000000000*mun + 10000000*dist + 100000*subdist + 1000*set

. format setor %15.0f

. list 

     +--------------------------------------------------+
     | uf   mun   dist   subdist   set            setor |
     |--------------------------------------------------|
  1. | 11    15      5         0     1   1.11500005e+15 |
     +--------------------------------------------------+

. format %23.0f setor

. l

     +----------------------------------------------------+
     | uf   mun   dist   subdist   set              setor |
     |----------------------------------------------------|
  1. | 11    15      5         0     1   1115000050001000 |
     +----------------------------------------------------+

[Clyde Schechter]

When you multiply uf = 11 by the first large number, you are already making a 16 digit number. If you try it with format %16.0f, it displays as an integer:

. format setor %16.0f

. list setor

     +------------------+
     |            setor |
     |------------------|
  1. | 1115000050001000 |
  2. | 1115000050001000 |
  3. | 1115000050001000 |
  4. | 1115000050001000 |
  5. | 1115000050001000 |
     |------------------|
  6. | 1115000050001000 |
  7. | 1115000050001000 |
  8. | 1115000050001000 |
  9. | 1115000050001000 |
 10. | 1115000050001000 |
     +------------------+

That said, unless you are going to use this "identifier" for calculations, given its length, you might be better off making it a string variable. See -egen, concat()- for how to put together a string out of those variables.

[Nick Cox]

I agree with Clyde's suggestion to consider string identifiers. Note also that from the outset (STB-50, 1999) egen, concat() was explicitly written to allow different punctuation, including spaces or commas or whatever. That would make it especially easily to take apart again, as often seems needed.