Count the lenght in numeric observations

[Daniel Perez]

Hello,

I am working with a numeric variable that, on some occasions, uses 10 digits, and 8 in others. I would like to count the occurrences of each situation. Any suggestion?

Thanks.

[Wouter Wakker]

If you have only digits, no decimals, you could do something like this.

clear
set obs 5
set seed 4321
gen x = runiformint(1, 2000)
gen length = strlen(string(x))
list
tab length


. list

     +---------------+
     |    x   length |
     |---------------|
  1. | 1808        4 |
  2. |  687        3 |
  3. | 1032        4 |
  4. |  328        3 |
  5. |   75        2 |
     +---------------+

. tab length

     length |      Freq.     Percent        Cum.
------------+-----------------------------------
          2 |          1       20.00       20.00
          3 |          2       40.00       60.00
          4 |          2       40.00      100.00
------------+-----------------------------------
      Total |          5      100.00

[Daniel Perez]

Excellent Wouter, this is what I needed.
Thank you!

[Nick Cox]

floor(log10(whatever)) should show 8 or 10 correspondingly.

[William Lisowski]

If the two choices are really just 8 digits or 10 digits

generate ten_dig = whatever >= 1000000000

And let me note that if you do have a numeric variable holding 10-digit values, it must be of type double or else you have precision problems and possibly incorrect values.

. set obs 1
number of observations (_N) was 0, now 1

. generate float x = 1234567890

. generate double y = 1234567890

. format %12.0f x y

. list

     +-------------------------+
     |          x            y |
     |-------------------------|
  1. | 1234567936   1234567890 |
     +-------------------------+

[Daniel Perez]

Thanks, Mr. Cox and Mr. Lisowski for your help!

[Nick Cox]

Also, 1 billion lies in between

count if x < 1e9 
count if inrange(x, 1e9, .)

[Laura Grant]

floor(log10(whatever)) should show 8 or 10 correspondingly.

This needs to be

floor(log10(whatever))+1

or it could be

ceil(log10(whatever))

Was looking up this to determine leading digit for Benford's law.