Recently, I had an inquiry from a former student of mine, now enrolled in a dissertation-level research methods class that was leaving her baffled. It seemed to me that she was being dropped into the middle of the great big methods ocean without much preparation or even basic vocabulary and structure. So I took a crack at setting out in a short form what I think ought to be the foundation for any such study. I’d like to share with you here what I gave her, and get your thoughts on its value.
We have to begin with some definitions: The distinctions among facts, theories, concepts, and variables are pretty critical to research. Basically, in any research task you are dealing with two different worlds, or levels: the level of experience (reality), and the level of theory (abstraction). Here’s a little diagram I used to use with my students to explain this.
The level of experience is where we live, and about which we want to make predictions. The whole idea of research is that if we can find out some things about how this world of ours works, we’ll be better able to make predictions and thus increase our own control of our destiny. So we want to stay rooted here. On the other hand, we want our predictions to be general, not just about how to find the next root. That requires us to think in terms of abstractions.
“Facts” are things that are observable in the real world using our regular sense-gathering tools – statements about reality. “Concepts” are ideas – the equivalent of facts, but in the level of abstraction. “Theories” are statements about relationships among concepts. And “variables” are real-world facts that we have “mapped” onto our concepts. They are things that we can measure here on earth that we believe (and can support our belief) to be the equivalent of a theoretical concept.
A simple example: suppose that we believe that worker satisfaction is a good thing, and leads to other good things. So we formulate a “thesis statement” in general conceptual terms: “worker satisfaction leads to improved productivity”. Note that neither of these things really exists. You can’t hold satisfaction in your hand, although you believe that it is a real phenomenon. Your thesis statement is made at the level of theory, or abstraction. Both satisfaction and productivity are concepts, and your thesis statement brings them together into a proposition.
But since neither of these things really exists, if we’re going to study their relationship we need to find real-world things that we can actually measure that we believe are credible analogues of our concept. So we start thinking – what might be indicators of satisfaction? Well, for one, we could actually ask the workers how satisfied they are with their work. There are reasonable arguments as to why this might be a less-than-perfect indicator, but it’s hard to argue that it is not relevant. Regarding productivity, the same process: one possible indicator might be the number of cases the worker is able to process.
Once we have the idea of how these concepts might be turned into measures (a process called “operational definition”), our next step is to identify the “variables” or things that we can actually measure with the tools we have. Satisfaction we can measure with a questionnaire to the workers: “how satisfied are you with your work?” We might give them a scale on which to respond, from “very satisfied” to “very dissatisfied”, with a few steps in between. We expect that there will be some variation in the answers among the members of the group we’ve decided to look at. Regarding productivity, we can simply count how many cases a given worker processes per hour; that’s probably already being tracked. So now we have identified two variables: “satisfaction score” and “cases per hour”, that we can actually measure and associate on our level of experience. We’re making the great big assumption that since these variables map well onto the theoretical concepts, any association that we can identify here on earth will then support the idea of an association between the concepts at the level of theory. We call the association at the level of theory a “proposition”; when we formulate it as a relationship between variables, we call it an “hypothesis”.
So what do we do now? Well, here’s the actual field research. We find a company willing to cooperate with us, go into it, administer a questionnaire to the workers about their satisfaction, get the records of their productivity, go back to our office and enter these data into our computer (it’s a lot easier than calculating things by hand), decide what statistical test would be appropriate to look for a relationship between these variables, calculate it, and decide whether or not it supports our hypothesis. If we decide that the relationship is significantly stronger than what might have been expected by chance alone, we believe that our hypothesis is supported.
Notice: Not proved, but supported. It’s the best we can do. Proof exists only in realms of abstraction like mathematics. All supported hypotheses remain subject to rejection if later evidence does not support them.