Measuring educational attainment: an interactive guide
How to measure the highest level of education a person has achieved, and how to choose between measures
Oatley, S. and Costello, P. (2027) ‘Measuring educational attainment’, in Encyclopedia of Measurement in Social Sciences, 2nd edn, Vol. 3, pp. 386–393. Elsevier. Read the chapter.
Educational attainment measures an output of education: the highest level a person has completed. It is used as a control variable and as an outcome in its own right. It is harder to measure than it looks, because education systems change over time and differ between countries: a 16-year-old in England sits GCSEs, one in Scotland sits National 5s, and in some countries there are no exams at that age at all.
The chapter’s four recommendations
- Avoid simple measures such as years of education.
- Within one country, use its National Qualification Framework.
- Across countries, run sensitivity analyses with more than one measure.
- Consider whether a scaled measure is practical for your study.
1Years of education
Years of education is easy to collect and can go straight into a model as a continuous variable. The chapter recommends against it unless nothing else is available, for three reasons: it confuses time spent with what was achieved, its implied ranking can contradict how the education system actually works, and the meaning of a year changes over time. Pick a scenario to see the problem.
Durations follow the chapter’s examples. Apprenticeships last from eight months to six years; the four-year example is one point in that range. Levels are from the NVQ table in section 3.
The third problem is credential inflation. In the 1970s about 60% of young people in England left education at 16, and five good O levels were a real achievement. By the early 2000s, university for half of all young people was a realistic goal. The same number of years meant something different. Section 6 returns to this.
2Highest qualification
Qualification-based measures use richer information: the type of course, academic or vocational track, and the grade. The simplest version asks people to pick their highest qualification from a list. Raw lists cause problems when qualifications and their names change over time, or differ between countries (the UK has both problems at once). The chapter advises against raw lists unless you know exactly how they were built.
Collapsing a long list into a few categories, even down to two, is sometimes advised for comparability. That throws information away. It is justified mainly by statistical power: a small sample spread over many categories (section 5).
3National Qualification Frameworks: the NVQ example
Most countries have a National Qualification Framework (NQF) that places academic, vocational and work-based qualifications on one ordered scale. In England this is the NVQ framework, a five-level ordinal scale. Its main strength is consistency over time: an O level and a GCSE never existed at the same time, but both sit at level 2. Its weakness is that comparisons across countries only work through a wider framework such as the European Qualifications Framework.
Source: chapter Table 1 (England). Routes: academic, vocational, and applied or work-based.
4Cross-national schemas: CASMIN and ISCED
For comparisons across countries, two schemas were designed for the job.
CASMIN
Developed in the 1970s for comparative social mobility research and refined since. Nine categories in three tiers (primary, secondary, tertiary), separating general and vocational tracks. Preferred in formal statistical modelling, and it performs better for European countries than for North American ones.
ISCED
UNESCO’s classification; the 2011 edition has nine levels (0–8) with detailed sub-categories. Mostly used for international reports. Its construct validity is low because countries interpret its levels differently, and surveys rarely collect the sub-categories.
| Tier | CASMIN | Description |
|---|---|---|
| Tertiary | 3b | Higher tertiary: completed academically oriented university education |
| 3a | Lower tertiary: shorter, vocationally oriented | |
| Secondary | 2c_gen / 2c_voc | Full maturity certificate, general or vocational |
| 2b / 2a | Intermediate general / intermediate vocational | |
| 1c | Basic vocational training above compulsory education | |
| Primary | 1b | General elementary education |
| 1a | Inadequately completed general education |
Source: chapter Table 2 (Brauns, Scherer and Steinmann, 2003).
Both schemas change model results compared with national frameworks. Any general schema sacrifices some national detail, and which countries lose most depends on the schema. Hence recommendation 3: with cross-national data, run the analysis with more than one measure and compare.
5Categories and sample size
Detailed schemas such as CASMIN and ISCED often leave some categories nearly empty even in large surveys. The chapter does not recommend collapsing them unless there is no alternative. This widget spreads a sample over schemas of different sizes. The shares are illustrative, chosen to be uneven as real distributions are; they are not survey estimates.
6Scaled measures
Scaling turns qualifications into a single metric score, using other information such as income or marriage patterns to rank them. Scales avoid sparse categories and awkward interactions, and they can adjust for credential inflation by scoring each qualification within its cohort. Their drawbacks: education has many dimensions and a scale squeezes them into one; it needs extra data; and few studies use scales for education, so they are less validated. If you use one, publish your documentation.
7Choosing a measure
8Check yourself
9Reading
- The chapter: Measuring educational attainment (doi: 10.1016/B978-0-443-26629-4.00256-2)
- Companion chapter: Measuring social class: an interactive guide
- Connelly, R., Gayle, V. and Lambert, P.S. (2016) A review of educational attainment measures for social survey research. Methodological Innovations, 9.
- Schneider, S.L. (2011) on measuring educational attainment across countries; UNESCO (2012) International Standard Classification of Education: ISCED 2011.