It's fitting that you mention Iraq. In the lead-up to war, a number of prominent realists, including Ken Waltz, John Mearsheimer, and Stephen Walt, warned that war with Iraq would be a serious mistake. Their beliefs were rooted in theory, broad qualitative empirical research, and case studies on Iraq. This "non-scientific" approach led to two generalizable and falsifiable claims: first, that sadistic leaders like Saddam could be deterred, and second, that offshore balancing is a promising strategy.
Consider the money, human lives, and international legitimacy that our government would have saved if it listened to these IR scholars. It would be enough to compensate for the $9 million we get per year from the NSF, I would say. Of course, not all political science is done correctly, and there are very problematic qualitative studies that do not lead to sound policy prescriptions. But that is why political scientists need more funding from the NSF, so that we can arrive at more precise, empirically verified conclusions and train graduate students with the best scientific tools to conduct their qualitative and quantitative research.