Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Calculating years of schooling

    I am trying to calculate total years of schooling based on year highest level of education was attained and the year of birth. From that I would like to calculate total years of schooling as follows:
    Year highest level of education was attained- year of birth- 4

    say variable x= Year highest level of education was attained

    y=year of birth
    I'm trying the stata command
    "generate yrschooling= x-y-4"

    But that doesn't work as I'm getting numbers from -59 to 61. Any ideas what the correct command should be?

    Thanks.




  • #2
    Sounds like you have missing data coded as numerical values not as missing values. Look at the results of tab1 x y and see if any of the values look "weird" or are labeled as missing. After that makes sure that the missing values are recognized by Stata as missing, see help missing values and help mvencode
    ---------------------------------
    Maarten L. Buis
    University of Konstanz
    Department of history and sociology
    box 40
    78457 Konstanz
    Germany
    http://www.maartenbuis.nl
    ---------------------------------

    Comment


    • #3
      Thanks. I tabbed and checked missing values and nothing looks weird or is labeled missing. I think the problem is with the formula I used:
      year highest education level was attained-yr of birth-4. If someone was born in 1960 for instance and decided to pursue a bachelors degree in 2010, though highly unlikely but possible, that would formula would give a response of 46, which doesn't make sense as years of schooling. If anyone has a better suggestion for how to generate a years of schooling variable from year of birth and year highest level of edu was attained would be great.

      Comment


      • #4
        I don't see how your formula would produce negative numbers if things are coded the way you say. In any event it seems like it would only work if people went to school nonstop without any years off. Are you sure there are no other education measures? It doesn't sound like you have the information you need.
        -------------------------------------------
        Richard Williams, Notre Dame Dept of Sociology
        Stata Version: 17.0 MP (2 processor)

        EMAIL: [email protected]
        WWW: https://www3.nd.edu/~rwilliam

        Comment


        • #5
          What had happened was that the year of highest level of education was attained went back to before 1940 so I dropped entries before 1947 (dataset is for 2012 to only look at the working-age population 16-65. Now I only have positive numbers, but yes the problem is as you mentioned Richard, that this assumes people only went to school. I could try year highest level of education was attained- year of birth- total yrs worked-4 but then again this would only correctly capture people in employment and omits people who were inactive or unemployed and it also gives negative numbers.
          Last edited by decipheringstata; 28 Nov 2014, 01:53.

          Comment


          • #6
            Do you have the actual diploma attained by the respondents?

            Why do you want years of education?
            ---------------------------------
            Maarten L. Buis
            University of Konstanz
            Department of history and sociology
            box 40
            78457 Konstanz
            Germany
            http://www.maartenbuis.nl
            ---------------------------------

            Comment


            • #7
              I have highest level of education attained and could derive years of education from there. But I wanted a more precise years of schooling measure which is why I was trying the above formulas. It looks like deriving years of schooling from highest level of education attained does seem at this point to be the most precise option. I want years in years to get mean years of schooling and to use it as an independent variable in regression analysis (though dummies could work for that as well).
              Last edited by decipheringstata; 28 Nov 2014, 02:19.

              Comment


              • #8
                See also answers in http://www.statalist.org/forums/foru...s-of-schooling

                Comment


                • #9
                  What is precise depends on what you want to measure exactly. For example, I would call years of schooling derived from highest level of education "pseudo-years of schooling", to make clear that "real" years of education and pseudo-years of education measure different things. Real years of schooling measures the amount of time you have spent in formal education. Pseudo-years of schooling measures the value of a diploma. The two are related but not the same. You can typically attain the same diploma with different years of education. Do you think it is the amount of time that your respondents are exposed to formal education that influences your outcome variable or do you think that it is the diploma attained that influences your outcome?


                  ---------------------------------
                  Maarten L. Buis
                  University of Konstanz
                  Department of history and sociology
                  box 40
                  78457 Konstanz
                  Germany
                  http://www.maartenbuis.nl
                  ---------------------------------

                  Comment


                  • #10
                    Thanks Maarten. I am actually trying to use both for comparative purposes though I personally believe that it's the educational level attained rather than years of schooling that is more relevant for the labour market outcomes in question. But what I'm trying to do is compare the two on the different outcomes. The tricky thing to creating what you call pseudo-years of schooling is that pre-primary school is not obligatory. so attaining ISCED 1 definitely requires six years of schooling, but it could be a cumulative of 8 if the person had done pre-primary school as well.
                    Also problematic is ISCED 5 because it could be anywhere between 4 and 10+ years. So how would you define the pseudo-years in that case? I tried the below:
                    2 yrs if ISCED==0

                    6 yrs if ISCED==1
                    10 yrs if ISCED==2
                    12 yrs if ISCED==3
                    13 yrs if ISCED==4
                    16 yrs if ISCED==5
                    Last edited by decipheringstata; 28 Nov 2014, 07:58.

                    Comment


                    • #11
                      "decipheringstata": I think you are on a Mission Impossible in the sense that, exactly as Maarten says regarding the two measures: they "are related but not the same". So, I don't think there is particular "crosswalk" that is ideal. Nonetheless, I concede that you're not the first person to try and do such crosswalks, especially when trying to compare, e.g., European data with USA data ("years of schooling" measures dominate in the latter). So, I recommend that you search the web for previous attempts to crosswalk, and benchmark your schema against theirs.

                      PS would you please re-register to use your real name (firstname lastname) on the Forum -- as the Forum FAQ points out, real names are strongly preferred here. [Names like "decipheringstata" are contrary to recommended practice (and may actually turn some people away from helping you).] It's easy to re-register: hit the Contact Us link at bottom of page, and make your request. Thank you.

                      Comment


                      • #12
                        I would be careful when assigning the same number of years to the same ISCED for different countries. Educational systems can differ quite a bit across countries (and within countries across provinces/states and across time). You could look at:

                        Heike Schröder and Harry B. G. Ganzeboom (2014) Measuring and Modelling Level of Education in European Societies. European Sociological Review 30 (1): 119-136.
                        http://dx.doi.org/10.1093/esr/jct026
                        ---------------------------------
                        Maarten L. Buis
                        University of Konstanz
                        Department of history and sociology
                        box 40
                        78457 Konstanz
                        Germany
                        http://www.maartenbuis.nl
                        ---------------------------------

                        Comment

                        Working...
                        X