Take a SERIOUS look at the UMich research track, 5-years for adult, 6 years for combined child and research. If there's somebody there you want to work with, and want to have significant time to do research prior to a post-doc (about 18 months protected time), it's a pretty sweet set up. The only reason I did not rank UMich #1 was because there really isn't anybody working in the relatively small field in which I'm interested.
WPIC has resources out its ears, and has a very flexible research track that provides around 10-12 months mostly protected research without adding any extra time to the residency, essentially putting you in a perfect position for your post-doc. It's the biggest tent in America, and whatever you want to do, there's probably someone there doing something related to it. Except PTSD, for some reason, so the residency director told me once. Maybe because Pitt is strangely detached from the nearby VA? In full disclosure, I ranked them #1.
No love for Duke? Very flexible, and plenty of resources. They propose a very strange "do a year of research for your pgy-1" optional curriculum that, although I can't imagine why anybody would want to do it, shows that they're at least VERY flexible in trying to make research careers happen.
Wash U is flexible with resources and scheduling, but is a smaller tent than UMich and a MUCH smaller tent than Pitt. Cincy has some of the best bipolar work in the country, and their research track is strongly supported, and gets you started earlier with research than just about anywhere (UM and Pitt both expect you to have a few years of clinical under your belt, while your first research month at Cincy will be PGY-1).
Hopkins is of course exceptional for research career, but I'm convinced they use their residents as cannon fodder. I don't know why anyone would go to Hopkins if they could go to any other program in this thread.
Of course, the coastal programs turn out researchers aggressively, and Penn is probably the most underrated in that respect compared to MGH and Columbia and even Yale. On the interview trail, it seemed like in general that the prestigious East coast programs were on average much less flexible with their curricula than other schools, and plenty of folks I talked to felt their clinical loads got in the way of them being able to take advantage of the pletheora of research resources. Yale's child research track is a pretty amazing curriculum, but their faculty interests seemed narrow and entirely unrelated to anything I was particularly interested in. Lots of autism and neuropsych research, hardly any affective disorders.
West coastally, Stanford treats a small number of its residents to a very well supported research track, but otherwise leaves the rest of its residents to languish, at least for those going into child. There was literally no way that anyone I talked to could come up with a way for me to have 5 minutes of protected research time if I stayed there for 5 years for adult and child training. They did have a mechanism by which folks could apply for funding for research time their 4th year if you were adult track. Stanford soured the whole West coast for me, so I cancelled UCLA, UCSF and UCSD interviews, though I imagine they're pretty amazing research institutions that others can comment upon. I did interview at UW for personal reasons, and while their resources weren't as profound as the UCs, I can't imagine anyone being unhappy there, as a researcher or a clinician.
Oh yeah, and I don't know anything about UTSW (except that my friend who is an intern there is very unhappy for un-program related reasons), but it probably shouldn't be left off of a list with good residency research opportunities.