Thanks for this note. I’ve spent some time talking with the IPUMS International Team about this and we all agree that these frequencies do not seem quite right.
The IPUMS International Team constructs EDATTAIN and EDUCIE from years of schooling in primary, secondary, vocational, and tertiary levels. The challenge is that people reporting tertiary may also report primary/secondary years of schooling. Also, people may only report tertiary or secondary even though they clearly also attained lower level of education. This complicates the programming for the harmonization of this variable. Based on our discussions, the IPUMS International Team re-designed the programming for EDATTAIN for IE1971A and IE1981A. This re-design will show up on the live IPUMS International website with the next data release.
Another complication is that the number of unknown cases varies across the source variables. For example, in Ireland 1981, tertiary has more unknown cases than secondary, while secondary has more unknown cases than primary. This is also true in 1971. The differences in unknown cases are not trivial and they could be an explanation to the relatively lower than expected educational attainment.
The new frequency distributions for educational attainment still look slightly low compared to other samples, specifically for secondary and post-secondary education. Unfortunately, this is a problem that is likely explained by the large number of unknown cases existing in the data and is not fixable by the IPUMS International Team. The IPUMS International Team will add a note about this in the comparability section for EDATTAIN Ireland.