Wow, a lot of good discussion here. I'm in a bit of a hurry, but I'd like to comment on a few things. First, I definitely agree that there are some issues with how results are presented in journals. This is a problem not only because of the pressure to publish in R1 institutes, but also because of how concise journals want scholars to write their articles. Pages are at a premium in journals, and a lot of things get cut out that could help decision making with vetting methodology. I'm a huge advocate of that statistical analyses in journals need to be given a "Criticism" section. We may never know that people are reporting significant ANOVA findings, but they never told us whether or not the groups were equivalent. It's even worse for regression, which requires a lot of vetting before you can be confident about its validity. I bet there are zillions of regression reports being passed off as valid, but have multicollinearity throughout them. Of course, I doubt anyone in power is gonna take seriously my, "We need regression criticism sections in journal articles" plea.
Yes, grant money may have political agendas attached to it. However, not all research completed at a university is grant-funded. Much of it is basic research, and is just related to the scholars' muses. With that said, I'll always automatically trust university research more than research that comes out of a politically-oriented think tank. The latter is almost always going to be slanted to a specific political leaning and it's often conducted by hacks, some who may have never even went to grad school. At the end of the day, however, regardless of where research is conducted, whether it's grant funded, and whether it has a political slant, you have to just vet the methodology, the theory, and the analysis, and judge it on its own terms.
rubadubdub, if you haven't, read the best policy book ever written, Policy Paradox by Deborah Stone. She talks about a lot of the things you are talking about.
I don't know if quanitative sociology is really falling by the wayside. I can only speak to my department, which is pretty much split down the middle between postpostivist and quantitative people and constructivist and qualitative people. As I said earlier, whereas psychology has established itself as a postpositivist field, sociology will probably forever be in an ontological existentialist crisis. But I think that's really a function of the scope of items of investigatory interest that permeate sociology. Coming from a psych background, I've always been an empirical, quanitative guy, but my dissertation is qualitative, because, well that's the paradigm I have to work from to answer my research question. Positivism and quantitative analysis is simply not equipped to deal with my question. With that said, I generally don't consider qualitative research "science," because, I consider science to be primarily interested in prediction and replication. Ethography and phenomenology really aren't interested in these things. That is in no way to denigrate those methods of inquiry. They are very valid epistemological tools for helping us understand the world. I just don't think of them as scientific.
Ironically, I'm actually an antipostivist, even though I'm a quantitative guy. I see the social world as existing as a pure social construction, but through that construction, different patterns can emerge. I'm also neither a humanist nor an anti-humanist on epistemological grounds. What I mean is that the question of whether or not a such thing as "human nature" exists is technically a falsifiable question. However, to answer that question, you'd practically have to raise people in Skinner boxes to control for all of the confounding variables, which raises both methodological and ethical issues.
Agreed that SPSS and Systat aren't really geared towards large data sets. Being ANOVA-based, they are geared more towards psychologists, who are probably dealing with n sizes that are on the relatively smaller scale. One thing I never liked about SPSS and Systat is the output page, which is rather unituitive. I like Stata, because I can just punch in commands down at the bottom and then have my output pop up right there in a nice, neat package. I also prefer Stata's graphical tools. It's a lot easier to look for heteroskedasticity in output and the such with Stata. Just my two cents on the matter. I've never used SAS, although I have used the programming part of SPSS before. Not a fan and totally unintuitive to anyone who has done programming in any other capacity.
. No need to apologize! Strongly held opinions based on experience and training from the opposite sides of the aisle on a topic usually results in interesting discussions, as long as everyone stays polite.
But a lot of basic research that doesn't seem "important," actually leads to some astouding insights.
