Removing the last three digits of a four-digit numeric variable

[Rosa Blau]

Hi Statalist,

I am working with a four-digit integer variable, ISCO08_1. I would like to simply drop the last three digits, and create a one-digit integer variable "ISCO". So far I have tried versions of egen with the cut function, but am generating nothing but missing values. My data and code are below. I am grateful for your advice.

* Example generated by -dataex-. To install: ssc install dataex
clear
input int ISCO08_1
3343
3115
3343
3132
5141
5322
3251
2341
7511
5230
4222
7127
2636
9321
9111
9622
9214
7223
3343
1211
3411
5311
4221
3434
9111
5411
2341
5153
5153
7411
3343
5153
5230
2636
9411
9111
4321
3251
8171
5222
2423
2529
8152
5322
9411
3132
3132
2641
5230
5222
4222
2221
3512
2212
2353
1112
5141
9412
2163
2412
2142
7543
5322
8183
5131
7126
5169
2611
1311
9214
2212
5312
2263
2164
2514
3322
3339
3313
1211
9412
7322
7422
3221
3313
5169
3221
3115
1330
2641
2636
2221
8183
4321
2221
9214
2341
9112
2622
2330
7212
end
label values ISCO08_1 ISCO08_1
[/CODE]

egen ISCO = cut(ISCO08_1), at(1)
(1842103 missing values generated)

. 
end of do-file

. codebook ISCO

----------------------------------------------------------------------------------------------------
ISCO                                                                                     (unlabeled)
----------------------------------------------------------------------------------------------------

                  type:  numeric (float)

                 range:  [.,.]                        units:  .
         unique values:  0                        missing .:  1,842,103/1,842,103

            tabulation:  Freq.  Value
                     1,842,103  .

[Luis Muelle]

Rosa:
Make a copy of your variable: gen bibi=ISC008_1 and then recode this newvar knowing
that your minimum value is 1112 (I prefer to begin with 0001 in any case)
recode bibi (1112/1999=1) (2000/2999=2)(3000/3999=3) (4000/4999=4)(5000/5999=5) ///
(6000/6999=6)(7000/7999=7) (8000/8999=8)(9000/9999=9)
Confirm with tab bibi
NB: No obs with code 6?
Hope this is practical !

[Andrew Musau]

 gen ISCO= real(substr(string(ISCO08_1), 1, 1))

[Nick Cox]

gen ISCO = floor(ISCO08_1/1000)

[Rosa Blau]

Thanks for your advice Louis, that was actually what I ended up doing. But I felt like there needed to be a more elegant solution.
Both Andrew & Nick's suggestions worked, thanks a lot for those!

[Rosa Blau]

Hi, now I am trying to remove the last four digits from an eight digit numeric variable, "SIC_codes". Dataex below:

* Example generated by -dataex-. To install: ssc install dataex
clear
input double SIC_codes
      -3
      -3
      -3
      -3
      -3
      -3
      -3
      -3
      -3
      -3
      -3
      -3
      -3
54110201
83619905
34950000
35410200
26790401
50230000
30890308
87480202
27520101
51930200
36120000
50510000
73730000
27520101
57120000
80820000
50850000
80690000
55310100
83220503
81119901
87410000
60210000
27310000
28420102
20539904
83610000
83999909
30699905
64110000
26790301
48320000
60220000
80620000
49110000
70110000
50830300
81119902
86210000
16110000
80520000
82210102
83220300
28139909
34929901
50510217
81119902
81119901
41110101
20239905
36690100
30860000
84120101
80620000
30890608
82210102
87420000
63319908
35350000
34410000
81110200
30890600
87429902
65529901
80829902
73891300
50840000
17310000
80620000
83220000
70110000
80110105
87110000
86110000
80620000
87310302
35359901
63310200
60620000
80110101
87310200
73810000
22829904
80990102
55119901
80110505
15410000
end

Clearly I have not understood the underlying logic of the real and floor gen options mentioned above, as I can't get them to work in this case (they both produce results other than simply removing the last four digits of the eight digit variable).

Any advice on how to overcome this roadblock are much appreciated!

Best,
Rosa

[Rosa Blau]

OK, so

 gen sic4 = floor(SIC_codes/10000)

DID work. Not sure what went wrong the previous times. Thanks again Nick.