3 Data and measurement
One of the key differences from ordinary thinking and thinking as a researcher is in the vocabulary used to talk about issues and develop analyses. Although some of this vocabulary may be tied to a specific kind of positivism, it actually applies across all methodological approaches. This vocabulary provides a vital structure for designing, carrying out and reporting on your research projects.
Unit of analysis
The unit of analysis, in essence, is the unit you are interested in studying. It can be surprisingly hard to determine what the appropriate unit of analysis is. It might be individual people, dyads (e.g. pairs of people such as couples or employee/supervisor), formal organizations, geographic units are some common examples. In conversation analysis, the unit can be the sentence or conversation; in content analysis it could be newspaper articles, advertisements or works of art. In data mining it can be the individual words that make up texts.
That’s the simple version, and plenty to think about. But often in the real world of research you have multiple units that relate to each other. The example that is probably easiest for Americans to understand is that students are nested in teachers who are nested in schools that are in districts and states. Or individual people live in neighborhoods. In these cases we may have concepts that cross units. For example, do characteristics of a place (such as the level of air pollution) have an impact on the health of individuals? Do teacher to teacher differences explain differences in student outcomes? And are teachers with different characteristics found in different kinds of schools? Advanced statistical techniques can help to manage this kind of complexity, but qualitative research can also be effective in uncovering and exploring these connections of units.
Variables
Another key thing to consider is what variables we are interested in. But what is a variable? This is a surprisingly hard concept to get your mind around. Many of us first heard the term variable in the context of a math class. We might have seen something like this:
x + 5 = y
There are (infinitely) many values of x and y. They can vary. Variation is what makes something a variable.
However if I give you this equaton
x + 5 = 7
there is only one correct value for x, x = 2. In that case x is not a variable. It is just standing in for the number 2.
One way to think about it that can also resonate is to contrast variables with constants. A variable takes on many values, a constant takes on a fixed value. Importantly for doing research, you need to consider what will be constant in your study (for example maybe your study is in one school, in which case the school is constant) and what varies (for example, the years of teaching experience of each of the teachers in the school). So being a variable or a constant is not an inherent characteristic of a concept, it is deeply connected to the choices made in the design of a research project.
This can get complicated fast, for example if you collect data from the same person once a year for three years there are things about them that may change, but there are other things that you may assume are constant.
Hypotheses and Questions
Research is seldom this clear cut, but often we will say that either research has a hypothesis or it has a guiding research question.
A hypothesis is a predictive statement about what you expect to find. It is based on theory and past research — calling it an “educated guess” as some do is misleading. A research hypothesis should minimally be about a unit of analysis, two variables, and the expected relationship between them. A hypothesis is something that is either true or not true, and if we can come up with a research design for it, we should be able to conduct empirical research to assess this. (Note: This is not about the statistical concept of hypothesis testing.)
A guiding research question is similar in that its purpose is to inform empirical research. But a question is not predictive. Instead it provides a framework for exploring the relationship between variables. Most often this is associated with qualitative research that is interested in understanding the experiences of people as they understand them, and thus it is a very deliberate stance against imposing an external understanding on their perspectives (though of course the idea that their meaning making is central is itself a theoretical stance). But this approach is also surprisingly important in highly technical analyses that attempt to extract underlying meaning. The variables in a guiding research question can be harder to pull out because they are implied.
In both cases it is essential that they be researchable. The process of developing a research idea into a viable research project is often one of moving from very abstract ideas (that may be slippery and hard to assess) to something more specific that can be investigated empirically.
Measurement
To actually be able to research a topic we need to put all these elements together and to consider how to measure the variables we are interested in on the units of analysis we are focused on.
For example, suppose we are interested in religiosity or the strength of religious belief. We would think about that differently if talking about individuals in the United States overall, individuals who are members of specific religious communities in the United States, the characteristics of a congregation or characteristics a geographic community. We would have to ask ourselves many questions about what we actually mean by religiosity in the context of our study and we would also want to look at past studies and what did and did not work for them. This process can broadly be called conceptualization. Then we would want to transform this into an actual measurement of some kind (one or more survey questions, behavioral measures, open ended questions or observation of events). This process can be called operationalization. The relationship between these two processes is dynamic as we will see.
As you do the reading (including the measuring inequality exercise) think about what units and variables your are interested in. The exercise is designed to do two things. First to get you thinking about how many possible ways there might be to operationalize the idea of inequality (and of course there are many more) as well as to give you a bit of background thinking about some descriptive statistics.
Read: