Most rational observers agree that there is a serious higher education bubble going on, and way more theses and dissertations are produced that is optimal (although there's been a strong move towards non-thesis option Masters degrees). But we're in the habit of encouragement here at the Chariot---it seems to be our area of comparative advantage, so today I'm offering two suitable topics, absolutely free. You don't even have to cite me or credit me in any way.
Topic 1: Do a study of the ending outcomes of children who are homeschooled of the various main flavors (traditional or unschooling) vs the various flavors of private schools vs public schooling. Control by the IQ of the parents (you can get this from military records or the longitudinal studies that Murray used back in the Bell Curve, another alternative would use SAT/ACT scores). This beats the hell out of just controlling by race/SES/education, and controlling by the parent's IQ gives you an indication on whether any of the modes of schooling might actually increase IQ, which after all isn't 100% nature although most of it is in the context of a high surplus modern society. Some interesting questions you could answer this way, besides the obvious, might be which modes of schooling work best as a function of IQ ranges. Do homeschool kids overperform their IQ? Do homeschool parents tend to be smarter than their SES/education/race would tend to imply?
I imagine you could walk this topic in any of a number of departments, and it'd probably be worth several journal articles.
Topic 2: It has been observed that the homicide rate presently would be a lot higher if we still had the medical capability of, say, the 1980s (this is to say a lot of people treated today and who make full recovery from aggravated assaults and the like would've been remanded to a pine box 20-30 years ago). Put meat on this skeleton. Using hospital and other medical aggregated records, compute a normalized homicide rate, normalized to 1900 medical technology or thereabouts, or the earliest date where your records of homicide and medicine are of good quality. Consider also likely criticisms: is there any evidence that people murdered today receive more injuries prior to actually expiring? Do those intend on murder automatically adjust the lethality of their attacks to compensate? Is the trauma medicine capability significantly unevenly distributed throughout the country? Enough to make county-level rate significantly impacted by this effect? Has the particular means of murder significantly shifted? Like the first topic, you could walk this in several different departments, although I'd say Criminology is less likely to make a hash of it than medicine, although you'll probably want someone with relevant medical expertise (preferably a very long history of such, like a guy nearing retirement age in a major hospital in a big city who has worked trauma medicine most of his career) as a co author on several of your papers.