To estimate probable year of educational attainment in stata

[Kehinde Atoloye]

I have a DHS data for countries in Sub-sahara Africa, downloaded from IPUMS. I am interested in the variable that measures educational attainment. The educational attainment was reported in the year of survey, which is not the year the educational level was attained. Is it possible to estimate the probable year an individual achieved a specific level of education at a specific age or range of age such that I will have a time series of years at which individuals may have achieved a level of education?

[Daniel Schaefer]

It's difficult to say without knowing more about the structure of the data and without domain knowledge regarding eduction in Sub-sahara Africa. If this were the US, you might be able to infer that most people will graduate from high school around 18. So you might take the age of the graduated individual and use that to find the year they graduated from high school assuming they graduated when they were 18.

[Clyde Schechter]

This depends on the degree of accuracy needed in that variable. It is true that 18 is the modal age at high school graduation in the US, but there are substantial numbers of 17 year olds and some 19 year olds. And ages both younger and older than that do occur, though infrequently.

I have no idea what happens in sub-Saharan Africa, but I suspect that there is also variation.

[Daniel Schaefer]

Absolutely! Depending on your data and with sufficient domain knowledge you could model some of that variation by using other variables to predict who is more likely to graduate at 17 and who is more likely to graduate at 19. Again, it depends on how accurate your predictions need to be.