Should the social sciences be like the natural sciences? Wilhelm Dilthey didn't think so. This early 19th and 20th century figure who went on to influence Martin Heidegger, Hans Georg Gadamer, and Paul Ricoeur contended that the concept of Verstehen is crucial in our interpretation of human thought and behavior. Verstehen literally means "understanding," and Dilthey believed that whereas we look for explanations of phenomena in the natural sciences, Understanding or Verstehen in Dilthey's technical use as applied to the social sciences means interpreting human behavior in view of generalizations made from descriptions of past or ongoing behavior and whatever judgments or practical rules we can derive from such behavior. Dilthey fundamentally believed that human beings were both historical creatures and creatures with complex agency, and both these assumptions make us track what count as empirical data much differently in the social sciences than in the natural sciences.
Whereas the natural sciences look for laws that govern phenomena, Dilthey did not think that we could be so lucky in understanding human beings. Think of laws in terms of counterfactuals. In Physics, say, if you know a particles' position and velocity, you can determine the particles' behavior. You can even account for changes in the behavior. Not so with humans, beyond whatever general principles it's possible to derive. You can wonder why Sam went to the company picnic and give some reasons why he decided on going instead of staying home but you can't know, apart from taking in a whole host of other countless factors, if he's going to go if he gets a stomachache. Maybe he will, maybe he won't. Part of what we think about whether Sam will go under different conditions has to do with what we think is appropriate or inappropriate in the circumstance, and these conditions of appropriateness range from the most local possible to the most global. As Dilthey explained in his book, Introduction to the Human Sciences, this tells us that our understanding of human behavior, as opposed to other kinds of behavior, is inherently normative, about what people ought to do in such-and-such a circumstance.
Yet we need not conclude from Dilthey's characterization of Understanding that it is something fundamentally anathema to naturalistic explanation, and he himself did not seem to believe as much. Pace the way in which the concept of Understanding may have been taken up post-Dilthey, it is possible to construe his treatise on the social sciences as continuous with the rest of naturalistic inquiry. The biologist-turned-philosopher Massimo Pigliucci, probably unknowingly, is in agreement with Dilthey. He has argued that the way we interpret human behavior is fundamentally irreducible to the way in which we investigate the natural sciences because, in addition to making use of normativity in the social sciences, we more particularly believe that there's such a thing as agency, that is, that people and other animals (and maybe robots too) act from thoughts and values. Pigliucci writes in his book, Nonsense on Stilts, "Suffice to say that free will is a way to label the complex decision-making processes in which the human brain engages, consciously as well as probably subconsciously," and Pigliucci suggests that this is as much a naturalistic way to do rational inquiry as anything else.
Another consideration worth remembering, lest we make too big a deal of the possible discontinuity between explanation and Understanding, is that what makes the social sciences real and reasonable endeavors just like the natural sciences is that people within particular fields generate hypotheses and check them against empirical data, which Pigliucci reminds does not only require experiments, since science "can be done with an intelligent use of observational evidence."
The common thread in all science is the ability to produce and test hypotheses based on systematically collected empirical data (via experiments or observations). How these hypotheses are generated, how exactly scientists go about testing them, and the degree of success we can expect from different sciences varies from science to science and from problem to problem.
Dilthey argues regarding the social sciences in general that the two considerations of paramount importance when thinking of human beings is their historical conditions and also the degree to which human agency confounds our potential understanding. Pigliucci, however, pushes further, arguing that when considering the sciences in general, we can easily see that historical considerations permeate several scientific fields and that the issue of human agency is one instance of the more general problem of complexity. Pigliucci informs us that what count as empirical data have a lot to do with the kinds of problems people are interested in solving within a field and the degree to which a field is both historical and complex.
[O]n the one hand, we have a continuum from completely historical (paleontology, astronomy) to partially historical (evolutionary biology, geology) to essentially ahistorical sciences (physics, chemistry)... On the other hand, we have a second continuum, from sciences that deal with simple, highly tractable systems where one can apply strong inferential methods (physics, chemistry) to sciences dealing with extremely complex objects, where statistical treatment is necessary and where the ability to explain and predict phenomena is much reduced (evolutionary biology, psychology).
This point of the Two Continua generalizes beyond fields that are generally accepted as social sciences, fields such as Psychology, Economics, Political Science, and Sociology, and into fields which have sometimes been considered separate from the social sciences and regarded as the Humanities, fields such as Literary Studies or Religious Studies. If what has preceded is true, then the extent to which these fields make use of empirical data and test hypotheses, they are sciences. Once we bear in mind a couple of the common features of the sciences, and how the Two Continua bear on the kind of empirical data the fields can collect, we have less worry to think that the social sciences are radically discontinuous from the natural sciences, and neither for that matter the Humanities. Human beings are both historical and rational creatures, and our empirical data follow from those bedrock assumptions, but then that's not any different from any other science, from which the research programs are built on their respective assumptions and on the basis of which questions constitute interesting problems in particular fields.