* 1 *
Sharpen Your Tools
Methods are means to an end, and you control them by using things in your brain, like concepts, which are mental tools. You donât want hand-me-down tools that are rusty with neglect. Know what you are doing and be willing to defend it.
What Are Methods of Sociological Research?
What Is Research?
This is a book about how to do sociological research. And so I think it is appropriate to talk about what research is. It is, first and foremost, work. One does not do work for nothing. Why do it? Even if you think it is only done for self-centered reasons (e.g., to get an A in a class, to get an article published, to get a job, or to get tenure), that begs the question of why there is a system in which this work is required.
The answer is that social science is one of the types of knowledge that require work to be done right, often very hard work, over long stretches of time, often on the part of many different people. Methods are about the how of this work. That means that if youâre not interested in methods, itâs like being a violinist and not being interested in playing. This is what you do all day . . . if youâre really doing something. And to get these methods to do something for us, we need to use them seriously, not ritualistically, and think them through.
How Do We Use Our Brains?
One of the things about our partition of sociology into theory, methods, and substance is that we forget that we can use our brains not just in the theory part but also in the methods part. In fact, the key argument of this book is that we need to âtheorizeâ our own practice . . . not in the sense that most folks mean by âtheorize,â which is basically to toss a lot of fancy abstractions around hither and thither. I mean the oppositeâwe need to really have a scientifically defensible understanding of what we are doing before we give much credence to the results. If you were a contaminant ecologist attempting to see whether fish in some lake had too many toxins, you wouldnât just stop by a fish fry some folks are having and take a nibble, would you? So why do the equivalent as a sociologist?
In order to understand what our dataâour experiences and interactions with the worldâmean for us, we need to spend a fair amount of time understanding âwhereâ in the world we are, and what we are really doing. Weâre going to be trying to think about both the planned and the unplanned aspects of research. Letâs start with the planned parts. The most straightforward way that we use our brains in research is to construct a research designâa plan for future research that guides our data collection efforts so that we can compile our data usefully, and that guides our data compilation efforts so that we can bring our data to bear on the questions weâre interested in. The key to research design is understanding which kinds of problems are most likely to be relevant to our proposed case of research and seeing if we can be clever enough to avoid a head-on collision with them.
A good research design can avoid many problems, but not all. Much of the difficult work comes in ways that our research design didnât anticipate or wasnât relevant for. For this reason, we will find that we need to think throughâcarefully, rigorously, and without mystification or wishful thinkingâall the steps that go into making our claims (also see Latour 2005, 133), starting from the very first: âHow did I get interested in this question? Where did I get the concepts I use to think about it?â And then on to: âHow did I end up at this particular site? Why am I talking to these people and not those?â Or, âWhy am I looking at these documents and not those?â And then on to things like: âWhen someone says, âYes, I approve of Obamaâs foreign policyâ to me, what is going on? What did I say to provoke this? What was the setting in which this was embedded?â and so on. If you do this, you have a remarkably good chance of doing something worthy of being called social science.
In other words, the process of producing knowledge can be understood impressionistically as a meeting between your mind and some part of the world (perhaps a particular person at a particular place saying a particular thing). To understand what this interaction produces, to make it truly a datum, as opposed to a profound mystery, requires reconstructing, as best as we can, the nature of the meeting. Figure 1 gives a schematic rendition of this process. The rectangle you are in represents all the âplacesâ (analogically speaking) you could have this interaction. The solid line represents your path, and the broken line that of the part of the world.
What is your path? Perhaps something very simple: you wanted to study young Americansâ attitudes toward sexual preferenceâdo they see it as genetic or not? Already, in a way, you have started moving down one path, merely by thinking of this question and not others, and in formulating the terms in which you are thinking about it (perhaps you are assuming that âbiologicalâ and âgeneticâ mean the same thing to people). But then you make other choices. First, you stayed in the city where your own school is, as opposed to going to one of the other twenty-five thousand towns and cities in the United States. Second, you got permission to pass out flyers in the lunchroom at two high schools, as opposed to the other two hundred schools in the area. And then you waited.
Some parts of the process happened behind the scenes, as far as you were concerned. Students got the flyers. Some immediately became paper airplanes. Others were the subject of guffaws in the cafeteria. Some were carefully folded and put in a back pocket, sometimes by a guffawer. Others were stuffed into textbooks. Some of those folded and stuffed flyers were only found months later. Some were found soon, and here and there, a student pondered whether to volunteer. More decided to volunteer than actually called you. Some called, but when you werenât there, they didnât leave a message; perhaps especially those without their own phones. Some made plans to meet with you but never showed. And when they did, you asked some particular questions (among others)âa few out of the near infinite number you could have asked. And only then did you get your (potential) information in response. This answer is only a datum when it can be placed, with great imprecision, of course, in this overall landscape of acts of choice and selection.
I will be emphasizing this selection and selectivity throughout. We often now may associate this issue with causal estimation. Thatâs only one teeny part of selectivity. Itâs about the choices that we make, and those that others make. We canât control othersâ choices, but we can theorize them. And we can control our own. So make good choices. To do this, we need to pursue our investigations with symmetry, with impartiality (sine ira et studio, as Weber would say), and without bad behavior on our side. Iâm going to be arguing that you need to really pursue this ideal, and not simply in some vague lip-service, recognize-itâs-best-but-not-plausible-for-mere-mortals way, but as in, when someone draws your attention to a lapse here, you fix it. Sure, smart aleck reader, I also read the philosophy of science, and I admit that I canât prove to students that this is necessary, if you are going to do valid social research. But students have proven it to me.
To orient ourselves, letâs back up, and see what weâre doing with this whole âmethodsâ thing.
Methods in General
There are some things that areâor should beâcommon to all sociological methods. First, sociological methods are, I believe, driven by a question. That might sound funny or obvious, but itâs not; in fact, most sociologists seem to disagree with me here. But I think that things we do that arenât driven by a question arenât methodsâtheyâre not a means to an intellectual end. Theyâre dressing, ritual, whatever.
Second, sociological methods involve a good faith attempt to find a fair sample of the universe at question. Not a ârepresentativeâ one, but a fair one. A question has multiple places where it can be answered, and your answer may depend on which place you examine. For example, say you start with the question, âDoes strong leadership increase or decrease membersâ attachment to the group?â You might get a different question if you looked at army platoons than if you looked at the history of the British monarchy.
There are two implications. The first is that if you have an answer you want to find, it isnât fair to choose a site thatâs likely to give you the answer you want. The second implication (which Iâll discuss more in the next chapter) is that if it really seems like your answer completely depends on where you look, you donât yet have a proper question.
Third, sociological methods push us to be systematic in answering the question, allowing for disconfirmation of our hunches as opposed to selectively marshaling the evidence we want. In the most satisfying cases, we construct a âresearch designâ like the mousetrap in that board gameâa whole elaborate machinery is set up, then we pull the switch (by collecting our data), and see what happens.
Itâs rare for research to unfold so mechanically. And for that reason, sometimes we need to be âunsystematically systematic.â That is, we have to figure out whatâs the sort of evidence that we havenât seen yet and that might lead to a different conclusion. (This is what Mitchell Duneier [2011] calls âinconvenience sampling.â)
Fourth, sociological methods tendâif only for rhetorical reasonsâto stress comparison and hence variation. Itâs hard to know if youâre right or wrong in explaining something that doesnât varyâsomething thatâs always there. So some of the most interesting questions get ignored. Questions like âWhy do people use language?â or âWhat causes patriarchy?â arenât ones we can do much with. Something that does vary, however, is amenable to explanation via comparison. Iâm not actually so sure a focus on comparison is always a good idea, but it is a central aspect of sociological methods, so we would do well to understand it.
Those four traits are basic to most methods. Past these commonalities, we will find that different methods have different advantages depending on, most important, what the process is (or was) that is of interest to us. Is it social-psychological? Institutional? A historical development? Second, which method is most appropriate will turn on whether the effects of this process are . . . repeatable or not? Generalizable to all people or only to a group? Conscious or not conscious (can you get the information by asking or must you watch)?
Depending on what we think weâre studying, weâre going to take a different approach. Andâsilly though it sounds, I knowâto know what we think weâre studying, weâre going to need to make a few provisional decisions about what we think is out there in the world.
Things and Concepts
What Is Real? What Can Act? What Is a Concept?
Theory is a funny thing. Among the tricks it can play on us is leading us to devote long periods of our lives to the examination of things that, in our saner moments, we would concede do not exist.
Do you want to argue about what ârealâ means and start a rumble over realism? I donât. All we need to do is to use the word ârealâ to denote the stuff weâre absolutely committed to defending. That means something that probably has most of the following properties. First, we think itâll still be there if we come back tomorrow (itâs âobdurateââthough we donât deny that some phenomena are transitory). Second, we think that other people will be able to see it too (itâs âintersubjectively validââthough we donât deny that sometimes you have to learn to notice certain phenomena). Third, you can study it via a number of different methods (itâs ârobust under triangulationââthough we recognize that sometimes we donât yet have ways of getting information on peopleâs thoughts other than talking to them, and so on). Something that lacks one of these properties might still deserve our commitment. But something that lacks all of themâit comes and goes, not everybody can see it, and only some methods reveal itâthat doesnât seem like the kind of thing we call âreal,â does it? It sounds more like a ghost.
Sometimes we end up chasing ghosts because we are enamored of a theory that makes strong claims about the world. The simplest way in which this messes with our heads comes from what is called âreificationâ or âthe fallacy of misplaced concreteness.â In our case, that means that we take a âtheoreticalâ phrase or a conceptâsomething that should really be a shorthand that helps us organize our thoughts about the worldâand treat it as if it were a thing. Once it becomes a thing, it is easy for us to imagine that it can âdo thingsâ that we include in our explanation. For example, many sociologists are interested in capitalism as a mode of production rooted in the private appropriation of productive capital. Even if we assume that there is a utility to this theoretical construct (which Iâll continue to use as an example below), it doesnât mean that capitalism is really a thing that exists somewhere.
There are times when it is going to help to decide, before you begin, what you think is real enough to make things happen. The reason is simpleâthatâs the stuff you need to get data on. Once weâve done that, we can begin to think methodologically. Iâm going to start laying out the conventional understanding of sociological methods. I donât think that this understanding is defensible in all aspects, but it is important that you understand it, and appreciate its strengths and limitations, before we go much further.
What Is a Unit of Analysis?
Let me start by quickly going through some pretty conventional language that weâll need. Most folks will tell you that any sociological investigation involves a choice of the unit of analysis (UOA). These days, we frequently do comparisons across more than one case (or instantiation) of the unit of analysis.1 We generally have a question about some sort of variation or differences among instantiations of our UOA. Here are examples: does labor unrest affect a nationâs employment insurance policies? Do cities with higher levels of income inequality have higher rates of crime? Are religiously orthodox individuals more or less hostile to immigration?
Very frequently UOAs are nested, in that there are distinct levels of analysis. For example, nations are composed of states or provinces, which are composed of counties, which include towns. And sometimes it helps to see all of these as composed of individuals. Because of this, we frequently refer to the question of the choice of UOA as a choice of the âlevelâ of analysis.
Obviously, the choice of UOA determines the research design. If youâre going to compare UOAs, for example, you will miss everything if your choice is wrong. For example, imagine that income inequality does lead to increased crime at the level of the nation but not at the level of the cityâsay, because the processes have to do with cultural senses of fairness and opportunity and not material opportunism. You could come up with a null finding if you compare cities within the United States. Obvious, true, but still the sort of obvious thing that we overlook as we rush to âget started.â
Finally, there may be theories or questions that we believe to hold at more than one UOA. For example, we may say that increasing income inequality leads to increased crime at any level of analysis: town, county, state, or nation, say. While there might be times when this makes sense, itâs usually a worrisome signal that weâre tossing words around somewhat randomly. Since itâs unlikely that the processes involved at these different levels could be the s...