Grouping educational level in different variables

[Fëanor]

hi there!

I'm trying to apply a model based on the Human Capital Theory.
Besides other variables, I have grouped the educational level (in years) in 4 distinct variables, in the following way:

1. E1: variable that assumes values 1, 2, 3 or 4 in the case of the worker have these educational levels. Assumes value 0 otherwise.
   
   E2: variable that assumes values 5, 6, 7 or 8 in the case of the worker have these educational levels. Assumes value 0 otherwise.
   
   E3: variable that assumes values 9, 10, 11 in the case of the worker have these educational levels. Assumes value 0 otherwise.
   
   E4: variable that assumes values 11 or above in the case of the worker have these educational levels. Assumes value 0 otherwise.

My question is: grouping educational level this way may cause any problems to my estimation?

[trubador]

Since you wish to use four categories of educational level (say edu), you need to generate 3 dummy variables:
E1 = 1, if edu = 1,2,3 or 4 and 0 otherwise
E2 = 1, if edu = 5,6,7 or 8 and 0 otherwise
E3 = 1, if edu = 9,10 or 11 and 0 otherwise
(0,0,0) will yield the fourth category by design.

Instead of creating nominal variables like that, you can use the natural order in education and generate a single variable in the ordinal scale.
E = 1, if edu = 1,2,3 or 4
E = 2, if edu = 5,6,7 or 8
E = 3, if edu = 9,10 or 11
E = 4, if edu = 12 or higher

[Fëanor]

Since you wish to use four categories of educational level (say edu), you need to generate 3 dummy variables:

Instead of creating nominal variables like that, you can use the natural order in education and generate a single variable in the ordinal scale.

I'm aware of these options, but what I want to do is something slightly different:

E1 = 1, 2, 3 or 4 if edu = 1,2,3 or 4
E2 = 5, 6, 7, or 8 if edu = 5,6,7 or 8
E3 = 9, 10 or 11 if edu = 9,10 or 11
E4 = 12 or higher if edu = 12 or higher

I know that this is something unusual, but what I want to know is if it may cause problems to my estimation.

Thanks!

[EViews Glenn]

Shouldn't cause any problems with estimation as long as you don't also include the original EDUC. Simply implies a different return to additional years of schooling in the different regimes.

Note that there are a couple of different ways to do the coding in EViews. Based on what you wrote:

Code: Select all

series e1 = @inlist(educ, "1 2 3 4")*educ series e2 = @inlist(educ, "5 6 7 8")*educ series e3 = @inlist(educ, "9 10 11")*educ series e4 = @recode(educ>=12, educ, 0)

The following should also work

Code: Select all

series e1 = @recode(educ<=4, educ, 0) series e2 = @recode(educ>4 and educ<=8, educ, 0) series e3 = @recode(educ>8 and educ<=11, educ, 0) series e4 = @recode(educ>11, educ, 0)

if the values of EDUC are only integers

[Fëanor]

Shouldn't cause any problems with estimation as long as you don't also include the original EDUC. Simply implies a different return to additional years of schooling in the different regimes.

No, I won't include the original EDUC.

I've thought that in doing this way I could get slope coefficients more precisely than just using dummies or only one variable covering all the years of education.

Note that there are a couple of different ways to do the coding in EViews. Based on what you wrote:

Code: Select all

series e1 = @inlist(educ, "1 2 3 4")*educ series e2 = @inlist(educ, "5 6 7 8")*educ series e3 = @inlist(educ, "9 10 11")*educ series e4 = @recode(educ>=12, educ, 0)

The following should also work

Code: Select all

series e1 = @recode(educ<=4, educ, 0) series e2 = @recode(educ>4 and educ<=8, educ, 0) series e3 = @recode(educ>8 and educ<=11, educ, 0) series e4 = @recode(educ>11, educ, 0)

if the values of EDUC are only integers

Thanks, man!