At the risk of annoyingly oversimplifying and omitting important exceptions,* I’d propose that three major efforts drive most of social science today:
1. Causal explanations of concrete human behaviors
People buy commodities at a given price, or vote for a given candidate, or even die at a given age. To explain why, social scientists often use either statistical models or controlled experiments. These two methods are conceptually related, because one interpretation of a regression model is that it mimics the results of an experiment.
Regression models are used across the social sciences, including such applied social sciences such as business and education. Economics is the discipline that historically has had the most influence on these methods, because economists tend to be good at math, the discipline is large and influential, and lots of concrete data on economic behavior is available. However, economists increasingly study all kinds of behavior that have nothing to do with money, and some sophisticated techniques for these purposes originate in other disciplines. For instance, education researchers developed Hierarchical Linear Modeling because they so often encounter individuals nested in classrooms, nested in schools, nested in communities. HLM is now used in other contexts as well.
2. Detection of unobserved psychological factors
Some important human characteristics are not concrete behaviors and are not directly observable. For instance, you can’t tell how much a teenager knows about US history by just looking at her. You can give her a 100-item test and compute a knowledge score from her answers, but much science and art goes into designing the test and interpreting the data. The same is true of emotional states, character traits, etc.
Once you have valid and reliable measures of such inner psychological states, you can put them into the kinds of causal models described in #1. But it is a major task just to determine who has which inner traits. By the way, if people know and can be trusted to disclose their own inner psychological states, then all this research is unnecessary. We’d just ask people whether they know US history, trust their teachers, or feel angry. An important premise is that we have unconscious or unarticulated inner lives that can be revealed better from outside. For instance, I’d find out how much US history I know by taking a test written by someone else.
Psychology–like economics, a large and influential discipline–has driven the development of these methods, but they are used across the social sciences.
3. Interpretation of purposive human activity in context
People’s behavior can (sometimes) be causally explained, but it also requires interpretation. Voting is a concrete act, but what does it mean for an American to vote in a church basement? (Note that this is not the same question as why some of our polling places are in churches. The causal explanation might have little bearing on the significance of this phenomenon.) Likewise, what are the meanings of a Balinese cockfight to the people who watch and participate–and, specifically, what does it mean when cockfighting is traditional yet illegal?
In ethnography, the emphasis is on interpretation, particularity, context, and translation rather than generalizable explanations or unconscious states. A characteristic method is to ask people what things mean to them in their most familiar settings.
Ethnography has–to me–an odd origin. Late-Victorian anthropologists wanted to turn traditionally philosophical questions about the nature of “Man” into empirical questions. They were Darwinians, so they presumed that our essential natures were evolved, clearly evident in prehistoric contexts, but obscured by subsequent cultural variation. So they visited so-called “prehistoric” communities to understand how they worked. Now most of that conceptual apparatus has been criticized. For instance, hunter-gatherer societies are in history, have often developed from other kinds of societies, and vary profoundly. But ethnographic techniques remain illuminating in all kinds of settings that no one would call “prehistoric,” including Silicon Valley office parks and even departments of anthropology. They are used across the social sciences, and they overlap with the humanities.
I haven’t mentioned a host of specific techniques or even whole disciplines, such as sociology and political science. But I’d propose that the three methodological programs described here are dominant. A field like political science takes its name from the phenomena it studies–government and politics–but it draws on, and contributes to, causal modeling, psychometrics, and ethnography. To the extent that all of these approaches make problematic assumptions (e.g., methodological individualism, or a simplistic fact/value distinction), then those assumptions are pervasive in the social sciences.
*e.g., Community Based Participatory Research, or historically-informed political theory, or research on social entities other than people.