The problem with Sociology as I see it is it requires a lot of inference on what the data means. And if the sample was too small, if the data wasn't gathered correctly....it's also frequently used as a weapon by the Left, and occasionally by the Right as well when the data fits their assumptions, and that puts any science in an awkward position because everyone has been robbed of their credibility, even researchers who never set out to make a political point in the first place. That said, there is an element of social justice in a lot of what Sociology studies, so when you're doing a study on the life expectancy of middle-aged black women living in poverty, or homeless street kids, your work ends up political regardless.
I actually worked for a Sociological Research outfit for a few years in college. I helped design and code the methodology for one study about memory, listened to probably 500? hours of recorded sociological interviews, a couple hundred pages of transcripts and coded a lot of the data.
And there's two distinct kinds of work done. One is very informal, and very personal, where a sociologist talks to as many people as they can and asks them all the same questions but lets them answer very freely. Studies based off data like that are highly interpretive and usually rely on samples that don't meet statistical muster. The other kind are hard data studies, surveys done by the thousands. Those do contain the kind of rigorous science and statistical analysis that actually supports conclusions.
So yeah, I was neck deep in Sociology for three years, and while I very much understand and agree with the "Feel" part of it, there is a lot of bonafide science being done. And the free response interviews aren't useless either, they provide a critical angle that large anonymous studies do not. It's just very slow data to work through and yes, it does lead to an emotional investment in the topic, because the researchers are actually talking to these people in person, sometimes for 20 hours or more over several sessions. It's the kind of data that you spend a life time collecting and building on. (And it's usually almost always supported with hard data as well.)