Standard educational attainment scale
August 19, 2018 6:37 AM Subscribe
For research on health, psych, sociology, econ, etc, is there a standard way of recording the subject's education level?
(This is not a research project, I am not a PI).
It seems like sometimes these studies ask for the highest degree attained: never finished high school, high school diploma, GED, Bachelors, Masters, PhD, etc. Other times they ask for the number of years of education. Both of these seem to have limitations: highest attained misses years of study, and simple years might misrepresent someone who dropped out and later got a degree through a non-traditional route. And asking both increases the complexity so that's probably not ideal either.
So for people who do research that collects that sort of data, is a consensus forming around "you really should collect educational attainment this way" ?
(This is not a research project, I am not a PI).
It seems like sometimes these studies ask for the highest degree attained: never finished high school, high school diploma, GED, Bachelors, Masters, PhD, etc. Other times they ask for the number of years of education. Both of these seem to have limitations: highest attained misses years of study, and simple years might misrepresent someone who dropped out and later got a degree through a non-traditional route. And asking both increases the complexity so that's probably not ideal either.
So for people who do research that collects that sort of data, is a consensus forming around "you really should collect educational attainment this way" ?
Years of education is preferred as it is continuous and easier to compare than categories.
posted by k8t at 8:31 AM on August 19, 2018 [1 favorite]
posted by k8t at 8:31 AM on August 19, 2018 [1 favorite]
My preference is for highest level obtained, but like you mentioned, there are limitations to any option you choose. Depending on the demographic you are working with, including an option for "some college" between high school graduation and bachelor's degree is nice and gives some additional refinement for the huge swath of the population that does have a high school diploma but does not have a bachelor's degree.
The US Census uses the following categories:
High School Graduate
Some College, No Degree
Associate's Degree-Occupational
Associate's Degree-Academic
Bachelor's Degree
Master's Degree
Professional Degree
Doctoral Degree
When I ask this question, I typically combine the two Associate's Degree options into one field called "Associate's Degree" and combine the Master's Degree, Professional Degree and Doctoral Degree into one field called "Master's Degree or Higher" and have been happy with that.
posted by mjcon at 8:45 AM on August 19, 2018 [4 favorites]
The US Census uses the following categories:
High School Graduate
Some College, No Degree
Associate's Degree-Occupational
Associate's Degree-Academic
Bachelor's Degree
Master's Degree
Professional Degree
Doctoral Degree
When I ask this question, I typically combine the two Associate's Degree options into one field called "Associate's Degree" and combine the Master's Degree, Professional Degree and Doctoral Degree into one field called "Master's Degree or Higher" and have been happy with that.
posted by mjcon at 8:45 AM on August 19, 2018 [4 favorites]
Years of education is nonsensical for making comparisons between groups with lots of education. Some people do a PhD in four years, while others may take ten. Or a masters can take 1-5 years. The ten yr people generally speaking do not have any more education than the four year folks, but this unmeaningful info will affect any numerical statistics. And you also have the problem that a 6 year BS counts the same as some (4+2) MSc, which is generally not what you want.
posted by SaltySalticid at 9:38 AM on August 19, 2018 [1 favorite]
posted by SaltySalticid at 9:38 AM on August 19, 2018 [1 favorite]
Like all good statistical questions, the answer is ... it depends.
At higher levels of education (bachelor's and above), years of study become progressively less meaningful, in part because education becomes much less linear. In that context, using categories also makes comparisons between groups a little more intuitive (e.g. the infamous "non-college-educated white men" we are informed voted for Trump and Brexit).
If you are dealing with global populations, though, years of education has a lot more force. For instance, according to this (scroll down for graphs), in 2010, Indians had a mean of 6.25 years of schooling, while neighboring Sri Lankans had 10.
posted by basalganglia at 10:31 AM on August 19, 2018 [1 favorite]
At higher levels of education (bachelor's and above), years of study become progressively less meaningful, in part because education becomes much less linear. In that context, using categories also makes comparisons between groups a little more intuitive (e.g. the infamous "non-college-educated white men" we are informed voted for Trump and Brexit).
If you are dealing with global populations, though, years of education has a lot more force. For instance, according to this (scroll down for graphs), in 2010, Indians had a mean of 6.25 years of schooling, while neighboring Sri Lankans had 10.
posted by basalganglia at 10:31 AM on August 19, 2018 [1 favorite]
Comparing education levels across countries adds another layer of complexity. Cross-country comparisons is often done through ISCED. For instance, the Programme for the International Assessment of Adult Competencies (PIAAC) is an international assessment designed to measure skills (literacy, numeracy, problem solving) of adults. Appendix 5 of the Technical Report of the Survey of Adult Skills (PIAAC), shows how ISCED is used to make education levels comparable across countries. In the PIAAC dataset, the variable on years of schooling was derived from the highest level of education.
posted by oceano at 11:57 AM on August 19, 2018 [1 favorite]
posted by oceano at 11:57 AM on August 19, 2018 [1 favorite]
Something I hinted at in my earlier comment (but did not elaborate much on) was the difference between the data collection and data reporting/analysis angle. Researchers doing analyses on a data set are limited to how the data are collected. In PIAAC, respondents were asked to report their highest level of schooling, and from that, the number of years of education was derived. So researchers using PIAAC data could use either measure in their analyses. However, it would probably be less accurate on average (for reasons mentioned in the comments above) to derive the highest education level from years of schooling.
From the data collection standpoint, there is a cost (e.g. time, resources) involved for all data collected. The longer it takes for respondents to answer the education level question, the less time there is to answer a different one. Questions that are "too long" or "too confusing" are more likely to be skipped. It's likely that there is some correlation between education level and skipping questions for being "too confusing." One way to shorten the educational attainment question, for instance, is to to combine the high school diploma and GED response options into one. (E.g. High school diploma/GED). Depending on the context of the survey, the target population, and sample size, the distinction between High school diploma and GED may not be meaningful at the reporting/ analysis level.
posted by oceano at 1:21 PM on August 19, 2018 [1 favorite]
From the data collection standpoint, there is a cost (e.g. time, resources) involved for all data collected. The longer it takes for respondents to answer the education level question, the less time there is to answer a different one. Questions that are "too long" or "too confusing" are more likely to be skipped. It's likely that there is some correlation between education level and skipping questions for being "too confusing." One way to shorten the educational attainment question, for instance, is to to combine the high school diploma and GED response options into one. (E.g. High school diploma/GED). Depending on the context of the survey, the target population, and sample size, the distinction between High school diploma and GED may not be meaningful at the reporting/ analysis level.
posted by oceano at 1:21 PM on August 19, 2018 [1 favorite]
It depends to some degree on what you are studying -- subsistence farmers in developing countries it may be best to ask if there has been any formal schooling at all, while if you're studying lawyers in advanced countries even the level of education may not be enough; the difference may be which school they graduated from. So the most important thing is horses for courses.
Educational attainment has a number of superior features in general, though.
It's clear -- I took a year of preschool, a year of kindergarden, a year-long a school-organized internship in university, several years as a part-time master's student (some of them enrolled as full time, some of them doing no work whatsoever on my degree) -- do some or all of those count? Like, I honestly couldn't tell you how many years of education I have, but I click "Master's" in a second.
It tracks results -- the application will say you need a high school diploma for a job, not to have spent 12 years in school; my thesis supervisor only had something like 16 years of education, but those included graduating high school early, graduating undergrad in 3 years and getting a doctorate at Cambridge, so he was more qualified than Joe Blow who dropped out of college after four years, even if they both had 16 under their belt.
It matches Census numbers (in the US and Canada at least) -- this allows you to do things like weight your sample to match the population or otherwise determine how representative it is.
If I was asking for general purposes in the US and didn't want the full set of categories (the Census bureau has 15), I'd stratify it less than high school / high school or GED / some college but less than a bachelor's / bachelor's degree / graduate degree. These groups have respectively, of people 25 and older, 22.5 million Americans, 62.5M, 57.6M, 46.2M, 27.8M. Or go with mjcon's 6 category group, breaking out the associate's degree holders (22.3M of the 57.6M with less than a bachelor's).
posted by Homeboy Trouble at 10:05 PM on August 19, 2018
Educational attainment has a number of superior features in general, though.
It's clear -- I took a year of preschool, a year of kindergarden, a year-long a school-organized internship in university, several years as a part-time master's student (some of them enrolled as full time, some of them doing no work whatsoever on my degree) -- do some or all of those count? Like, I honestly couldn't tell you how many years of education I have, but I click "Master's" in a second.
It tracks results -- the application will say you need a high school diploma for a job, not to have spent 12 years in school; my thesis supervisor only had something like 16 years of education, but those included graduating high school early, graduating undergrad in 3 years and getting a doctorate at Cambridge, so he was more qualified than Joe Blow who dropped out of college after four years, even if they both had 16 under their belt.
It matches Census numbers (in the US and Canada at least) -- this allows you to do things like weight your sample to match the population or otherwise determine how representative it is.
If I was asking for general purposes in the US and didn't want the full set of categories (the Census bureau has 15), I'd stratify it less than high school / high school or GED / some college but less than a bachelor's / bachelor's degree / graduate degree. These groups have respectively, of people 25 and older, 22.5 million Americans, 62.5M, 57.6M, 46.2M, 27.8M. Or go with mjcon's 6 category group, breaking out the associate's degree holders (22.3M of the 57.6M with less than a bachelor's).
posted by Homeboy Trouble at 10:05 PM on August 19, 2018
I work in a lot of different contexts - from highly industrialized cities, where someone may have 25+ years of education, to places where an adult may just be finishing primary school. For all, I lean towards measuring attainment, rather than years. How I do so depends on context. In some settings, I will include "some" for all (e.g. "some secondary / high school", "GED/high school diploma", "some college / technical school"). In others, I only ask highest portion completed.
This is a long way of saying: there is no standard; you choose the measure that fits your context; if you want to compare with other data (in the US, Census is obviously the big one, and NHANES is synchronized), copy their standard. Here's their info.
US HHS has standard measures for several things (race, ethnicity, sex), but they copy Census for ed.
posted by quadrilaterals at 11:17 PM on August 19, 2018
This is a long way of saying: there is no standard; you choose the measure that fits your context; if you want to compare with other data (in the US, Census is obviously the big one, and NHANES is synchronized), copy their standard. Here's their info.
US HHS has standard measures for several things (race, ethnicity, sex), but they copy Census for ed.
posted by quadrilaterals at 11:17 PM on August 19, 2018
This thread is closed to new comments.
Years makes more sense in contexts like less developed countries where the primary question isn't going to be whether you have a BA or not but rather, for most of the population, how many years of primary+secondary education you completed before dropping out.
posted by GCU Sweet and Full of Grace at 7:14 AM on August 19, 2018