Standardized Education Recoding: Methods

Concept

 The concept for the standardized variable is based on the international standard classification of education from UNESCO ;  ISCED97 (see references below), and uses highest attained level of education.

Only completed training is taken into account. If the original survey contains categories like 'started, incomplete, without leaving certificate', etc. than the standardized variable is coded to the next lower level.

LIS standardized levels

The levels are expressed in low, medium, high, and correspond to ISCED97 in the following manner :


1. LOW
, including
 

ISCED levels 1 and 2
  no education 0
  pre-primary 0
  primary 1
  lower secondary education 2A
  compulsory education 2A
  initial vocational education 2B, 2C

2. MEDIUM
, including
 
ISCED levels 3 and 4
  upper secondary general education 3A
  basic vocational education 3B, 3C
  secondary vocational education   4B, 4C
  post-secondary education
(including either shorter vocational courses
 or programs preparing for courses on tertiary level)
4A

3. HIGH
, including
 
ISCED levels 5 and 6
  specialized vocational education 5B
  university/college education 5A
  (post)-doctorate and equivalent degrees 6

9. All else,
including
 
 
  missing  
  unknown  
 

level undefined

 
 

other school degree

 
 

not asked

 
 

still in school

 
  These vague labels are kept in a separate category to keep the true levels free from 'pollution'

Recoding

The recoding was done using the variables PEDUC or PTOCC or both, depending on availability. The ISCED97 manual served as the main guide for all OECD countries. For other countries, the LIS contact person was consulted. In addition, we made use of several (web)-documents to guarantee the recoding to be as comparable as possible.

LIS tried to derive the standardized variable for as many datasets as possible.

However, it will be clear that standardization may fail for several reasons. For instance, when the survey did not collect levels of education, but 'years in education', or 'age when leaving full-time education', standardization could not be carried out.

A problem that we regularly faced is when original categories contain a mix of educational levels. If a value of PEDUC could correspond to, say both ISCED levels 2 and 3 and therefore both to low and medium, the enrolment figures from ISCED were used to see if the vast majority could be assigned to either of the two.

As a rule, successful recoding only resulted in clear situations. In other words: we did not allow ourselves any guessing.

The details on the recode are available per dataset, either giving the concordance table, or explaining why the recode was not carried out.

For those datasets where no recoding is made available, LIS users may of course feel free to use their own judgment and pursue a recode at their own responsibility.

 References

Classifying Educational Programs, Manual for ISCED 97, Implementation in OECD Countries 1999 Edition.  Paris: UNESCO.

 Education at a Glance: OECD Indicators, 2005 Edition , Paris: OECD.

 UNESCO, World Education Forum, Education for all , Paris, UNESCO

 Link to UNESCO ISCED97 classification :

http://www.unesco.org/education/information/nfsunesco/doc/isced_1997.htm

 

Copyright (c) 2011 Luxembourg Income Study all rights reserved