Today I had a meeting with my dissertation chair about another paper I'm working on (the one that was panned for being too long when it wasn't) and it turned into a meeting about my dissertation. He was really pushing me to conceptualize it as an example of a specific theory - as an example of how the world works that just happens to be taken from education - rather than as a dissertation about education.
In the good old days before I entered the Ivory Tower, I would have said all research questions were created equal. Oh, curing cancer might beat curing bruises from a useful standpoint, but "finding info about the world and how it works" was all intellectually equal. Right, not so much. Facts about the world, no matter how useful, that are isolated facts and not part of a system, just don't rank as high. That's why "practical" disciplines such as education have less prestige than "pure" disciplines.
So if you're doing research in education, you can conceptualize your question several ways.
One, it is a practical question of "does X work?". Education gets browbeaten for asking these atheoretical questions, but they have their place. Let's say one of your state senators decides that in order to improve education, all class sizes need to be 10 students or less. If his bill passes, the state is going to spend a lot of benjamins to reduce class sizes. So asking simply, "Do smaller classes lead to higher test scores/higher graduation rates/etc.?" without any kind of theoretical grounding is a reasonable question. One reason to criticize this work is that it doesn't lead to any guidance on how to answer follow-up questions. If you find that reducing classes sizes from 20 to 10 does make a difference, would it also be equally good to add a teacher aide to each class of 20? Would reducing class sizes to 15 have the same effect? What about lengthening the school year? Does it matter for all grades?
Thus, you can introduce theory. In alumni giving research, some folks draw on psychological theories of attachment. Alumni wish to identify with a school they perceive as successful, thereby boosting their own projection of success. Your hypothesize that alumni increase their giving if their alma mater rises in the U.S. News rankings. And if you find out it does, you can go on to test whether making more PR noise about your rise in rankings further improves giving. Great, now you're grounded in theory. But your work doesn't feed back into the work on loyalty and attachment. It's pitched at others in education, not to psychologists. Your central question is about education fundraising, not about psychology.
Your final alternative is to take a theoretical proposition and use education as a test case. Suppose you are interested in stratification. In a nutshell, stratification theory posits that we all want our kids to have at least as much status as ourselves. The trouble is, the elite have more resources to use as inputs on their childrens' behalves. Thus, no matter what the rules of the game are, those who already have win. So you might posit that if education is a pathway to the elite, those who are already in the elite will set the educational bar higher for the next generation. If a college education used to be enough to enter the elite, ambitious non-elites would start to get college degrees. But now the elite is getting bigger; maybe a masters degree will come to be seen as a minimum. Education is your test case here; your next project may well be on how the elite controls the Social Register.
Even at a really excellent education department, like, oh, Vanderbilt, most of the work being done falls into the second category. At the end of the day, it's about improving education. I offer no criticism of this, and in fact I think it's what "the public" expects from an ed school. We in LPO tend to frown on the first category - my advisor made it clear long ago that my dissertation WOULD have a theoretical grounding. There is no or else.
So why was my chair pushing me towards category three? In part, of course, that's because his own dissertation was like that. It was fundamentally asking a question about a theory of political science. I don't think that's the only reason, mind you, but he didn't give a list of reasons. He did go further and suggested actively insinuating myself into the networks of where those kinds of questions are being answered. As my advisor pointed out, this would give me a whole other set of peers. (On the other hand, it doesn't really open up any more jobs. It's still a PhD in ed, and you don't get into a sociology department with that.) I am finding myself that most of the work I think is interesting is Category III work, and it's being done by people I don't know at conferences I don't go to. That alone is an incentive for me to follow his advice.