replacing non-numeric values for numeric values in dataset

[Tom regt]

I have a dataset that contains the educational level(educ) for individuals that are in a certain industry(indnaics) for the years 2000-2014.
The educ variable contains non-numeric values see below:

 
year
educ
indnaics

2000
5+ years
6214

2000
4 years
6111

2000
5+ years
3399ZM

2000
2 years
622

2000
Grade 11
446Z

2000
1 year o
611M1

2000
5+ years
611M1

2000
Grade 12
493

2000
Grade 12
4451

2000
5+ years
6211

2000
4 years
712

2000
Grade 12
6111

2000
Grade 12
722Z

2000
Grade 11
6231

2000
Grade 12
813M

2000
Grade 12
5133Z

2000
1 year o
6231

2000
1 year o
334M2

2000
Grade 10
334M2

2000
4 years
5416

2000
5+ years
611M1

2000
4 years
611M1

2000
Grade 9
6241

2000
4 years
722Z

2000
Grade 12
23

2000
Grade 11
3MS

2000
4 years
713Z

2000
Grade 12
6212


 etc.

I would like to replace these non-numeric values(educ) for numeric values(eduyears) according to the following scheme:

 
Educational attainment
eduYears

N/A or no schooling
0

Nursery school to grade 4
4

Grade 5, 6, 7, or 8
8

Grade 9
9

Grade 10
10

Grade 11
11

Grade 12
12

1 year of college
13

2 years of college
14

3 years of college
15

4 years of college
16

5+ years of college
17

Is there any efficient way how I could replace this?

I have looked at using the replace command or using loops, but I can not seem to understand how I should do it. It could also be that I am looking in the wrong direction.
Is there anyone who could help me out?

Kind regards,
Tom

[Nick Cox]

help encode

Specifically, define a set of value labels first and then invoke them in an encode statement.

[Tom regt]

Thank you very much for your help.
However, I do not seem to come to the right outcome. In my previous question I have been stating wrong that educ was a non-numerical value as i just noticed that it is being considered as a numerical value(It is colored blue in data editor).

I tried to use the recode command to replace the values, but it did not work.
I am not sure what I did wrong.

These are the steps I undertook:

generate EduYears = .
(21992180 missing values generated)

recode EduYears = 9 if educ=="Grade 9"

I then got the following error

[P]     error . . . . . . . . . . . . . . . . . . . . . . . .  Return code 109
        type mismatch;
        In an expression, you attempted to combine a string and numeric
        subexpression in a logically impossible way.  For instance, you
        attempted to subtract a string from a number or you attempted
        to take the substring of a number.

I know that i'm doing something seriously wrong here, but i cannot figure out what.
Could you help me?

Kind regards,
Tom

[Clyde Schechter]

The first problem is with the -if educ == "Grade 9"- clause. You state that educ is in fact a numeric variable. So it may have a value label "Grade 9", but "Grade 9" is not a possible value for the underlying numeric variable. So you need to identify the numeric value that is labeled "Grade 9". That numeric value is what Stata actually stores in active memory and is what it will try to compare educ to in each observation. To do this, you first need to know the name of the value label associated with educ. That will show up in the fourth column of Stata's output if you run -des educ-. Then you can -label list- whatever that label is, and see what the numeric value for "Grade 9" is.

(If you want to get fancier, you don't have to actually run -label list- and find the value; you can also change the -if- clause to -if educ == "Grade 9":lblname- where you replace lblname by the actual name of the label. And yes, you can even side-step that with a macro extended function, but I think you ought to stick with the basics here until you are more comfortable with Stata coding.)

Also, you want -replace-, not -recode- here. They are related but distinct operations, and here -replace- is more congenial.

[Tom regt]

Thank you very much, this has helped me a lot!
Kind regards,
Tom