Conceptualizing education Harmonizing Education Level: single - PDF document

Conceptualizing education Harmonizing Education • Level: single hierarchy of competencies Measures in the European Social obtained. Survey (2004) • Duration: time spent in education. • The two are identical if: – Single, comprehensive grade system Harry B.G. Ganzeboom – No grade repetition Jürgen Hoffmeyer-Zlotnik • In fact, this never occurs in reality. SMABS/EAM, Budapest, July 3rd 2006 Harmonizing Education in ESS 1 Harmonizing Education in ESS 2 2004 2004 Problems with Duration measures Good things about Duration • Not equivalent to level measures in "divided" • One single, simple (?) question. systems; the correlation varies between systems: in some systems the correlation could • In all systems there is some correlation even be negative. between duration and level. • Question formulation: "American interpretation" • Simple, cross-nationally comparable of "years" actually refers to level. • Accounting problems: metric; ratio level, detailed. – What to do with part-time education • Has a nice interpretation in human capital – What to do with grade repetition accounting. – What to do with different starting ages – What to do with recurrent education – Special education Harmonizing Education in ESS 3 Harmonizing Education in ESS 4 2004 2004 Problems with Level measures Good things about Level measures • Hierarchy and metric need to be • However you construct a level measure, constructed using (A) judgement or (B) the consistency between judges, or criterion variables. between criterion variables is impressive. • Hard to compare between countries (and • The predictive power of education is within countries over time). (primarily) in level, not in duration. Harmonizing Education in ESS 5 Harmonizing Education in ESS 6 2004 2004 1

Our aims ISCED • Construct a cross-nationally (and • International Standard Classification of historically?) valid and comparable indec Education (OECD). for level of education. • Comparative measure of level of – Establish valid singular hierachies for each education country at multiple time points. – Seven levels – Construct a compative metric. – Documented (in small print) for some 25 • Assess measurement error of the OECD countries for their education system in constructed measure using data in ESS. 1997. Harmonizing Education in ESS 7 Harmonizing Education in ESS 8 2004 2004 ISCED – problems Education in ESS • Researchers cannot consistently apply ISCED97 to their • Respondent: data. – Showcard with national categories (except 7 – Ambiguity about those currently enrolled. countries). – Ambiguity about post-secondary, non-tertiary education (ISCED – National categories are recoded to ISCED by local 4). researchers; documentation preserved and included – No differentiation within tertiary (ISCED 5/6). in the data (except for Germany). • Too condensed. In many countries only 3-4 categories – Independent question on duration. are effectively used and some of these are very large (> 50%). • Father, mother, partner: • Not sensitive to divided systems. – Showcard with national or ISCED categories. • Not sensitive to historical variation. – National categories (if applied) are not preserved in the data, only ISCED is provided. • 'Common denominator' approach. Harmonizing Education in ESS 9 Harmonizing Education in ESS 10 2004 2004 Problems in ESS education Comparing different measures • National categories tend to be replaced by • Five measures: – OPT1: Detailed (local) categories, optimally scaled within ISCED categories in data collection; some countries. countries have not used more categories or local – OPT2: ISCED categories, optimally scaled within countries. specialties. – OPT3: ISCED categories optimally scaled within countries, but constrained between father, mother, spouse, respondent. • Different treatment of respondent and the three – OPT4: ISCED categories optimally scaled between countries, others, both in data collection and data father, mother, spouse, respondent documentation. – ISCED: linear interpretation (0..6) of ISCED categories. • Researchers obviously have given their own • Note that the these five measures are nested. interpretation to the ISCED categories. • Alternative measure: – DURATION in years, truncated at 23 years of education. Harmonizing Education in ESS 11 Harmonizing Education in ESS 12 2004 2004 2

Design SPOUSE FATH EDUC EDUC • Develop optimal scores: first approximation is effect- LEVEL proportional scoring for a composite of criterion EDUC variables: father‘s and mother‘s education, spouse‘s MOTH education (FISCED, MISCED, PISCED), occupation SPOUSE EDUC OCCUP (ISEI), spouse‘s occupation (PISEI). • Use duration (DUR) as an alternative source of information (independent measurement and measurement error!). OCCUP • Assess loss of quality of measument in a multiple indicator status attainment model, estimated as a simultaneous equations model [SEM]. OPT1 OPT2 ISCED DURAT Details Isced Linear xnat local local xnat Harmonizing Education in ESS 13 Harmonizing Education in ESS 14 2004 2004 Expectations on measurement Problems with this approach relationships • Criteria • Circularity: we use the optimal scores, generated by the same data. – Loss of measurement quality is expressed relative to – Potential solution: estimate optimal scores in ESS02 and apply OPT1 (reference measurement relationship). in ESS04. – Parallel measurement makes for a true score model. • We currently use a primitive method of generating • Expectations: optimal scores. – Quality of measurement will decrease as we allow – Solution: use Princals or LEM. fewer degrees of freedom. • Design assumes – Duration will be a much worse measure of the true – single hierarchy, score than any the level measures. – hierarchy is the same for father, mother, spouse, respondent, – hierarchy is the same for ED/OCC, FED/MED, FED/ED. – Duration measure will not work as well in divided systems. • We have not done the SEM models yet. Harmonizing Education in ESS 15 Harmonizing Education in ESS 16 2004 2004 Data in ESS04 Special case: Germany • Germany claims to have a uniquely complicated • 24 countries to start with. education system. • 2 countries dropped because of missing or • So they ask two questions: one on academic training and on vocational training. invalid education data (GB, PT) • Hoffmeyer & Warner (2005) show how these two • 4 countries do not have a local measure of questions can best be combined in one single hierarchy with 10/12 levels. Not used by ESS researchers. education (AT, FI, IS, SI) and 2 other have local • Germany is the only country in the ESS that used a measure identical to ISCED (IR, UA). many-to-many mapping between local education measure and ISCED (measures are not nested). • 18 countries with all relevant data. • Correlation FED/MED (0,45) is relatively low to other • Only men & women with valid occupation codes countries (0.65) and other German data. Something went wrong! (whether currently employed or not), N=29057. Harmonizing Education in ESS 17 Harmonizing Education in ESS 18 2004 2004 3