Chapter 1 Variables and Their Measurement
1.2 Concepts, Measurement, and Operationalization
You might be wondering why we even need to introduce a concept such as variables. Can’t we simply call them characteristics, if that’s what they are? The short answer is that we use the language of variables when we engage in formal research, but the reason is not solely scientific jargon. Variables, as opposed to characteristics, imply measurement.
You see, sociologists and other social scientists study concepts (i.e., ideas, notions) that are more often than not abstract. If I say “I want to know if the average height of Canadians has changed over time”, it’s easy for you to suggest that I first collect information about people’s heights (perhaps actually measure them, if I don’t trust self-reports). By doing that, you might not realize it but what you have done is actually offer a way to measure a concept, which is what we call with the mouthful of a word operationalization. In other words, you have operationalized the abstract concept (height of Canadians) through the actual, physical measurement of individuals’ heights (in centimeters or in inches) in real life.
So operationalization is that easy, right? Unfortunately, no, not really.
What if, instead of average height of Canadians, I had wanted to study how poverty has changed in Canada over time? Or homelessness? How about income? Or people’s attitudes to immigration? Or their religiosity? What about if I wanted to study self-esteem of adolescents? Or social status among Canadian university students? Or bullying in high school?
I’m sure you have no trouble understanding the concepts as abstract ideas — but how do you measure them? [1]
Do it! 1.1 Measuring Homelessness
Imagine you really do want to study the prevalence of homelessness in your city (or any of the abstract concepts mentioned above). Before you decide how to collect information about it, you have to choose about what exactly you will be gathering information. How are you going to define being homeless in order to measure homelessness? In a word, how are you going to operationalize homelessness? Make a list of possible definitions. What are the various aspects of homelessness, which you may choose to consider in your definition or not, that make defining homelessness difficult?
All in all, operationalizing a concept boils down to choosing a working (i.e., operational), measurable definition of a concept within a given study. Most concepts can be (and regularly are) defined differently by different researchers. What matters is that the definition of any concept is provided and is used consistently within each individual study.
If the Do It! exercise above seems too abstract still, perhaps one easier way to understand the operationalization of concepts into measurable variables with concrete definitions is to imagine a survey question about what you want to study. Sometimes one such question can provide the operationalization/definition of the concept under study. Other times a single question is not enough and a set of questions can help a researcher measure what they want to study.
Let’s say you want to study income (perhaps as a part of a larger study on poverty). You want to ask people about their income but how exactly? Will you be asking about personal or household income? Are there types of income you have in mind — from salary, from rent, from interest, etc.? Is it weekly, bi-monthly, or annual income? Is it income before or after taxes? For that matter, do you mean only taxable income? Furthermore, what kind of answers would you accept? Will the respondents provide an exact number? Or will you provide a set of multiple-choice answers from which the respondents will choose?
For example, you can measure income in a hypothetical study (through a survey question) like this: “What is your household’s annual after-tax income (from any source)?” This means that you have chosen to operationalize the abstract concept income through the specific, measurable variable annual household after-tax income.
The types of possible answers you choose to accept for the question are also part of the measurement. Example 1.1. below offers three options to operationalize income.
Example 1.1. Operationalizing Income
Q1. What is your household’s annual after-tax income (from any source)?
a) $0 – $50,000;
b) $50,001 – $100,000;
c) $100,001 – $150,000;
d) $150,001 – $200,000;
e) $200,001 or more;
Q2. Is your annual household after-tax income (from any source) less than $50,000?
a) Yes;
b) No.
Q3. What is your annual household after-tax income (from any source)?
…. [Any number provided by the respondent will be recorded.]
The multiple choices provided in Q1 in the example above can contain any number of categories to choose from. I have chosen to go by 50 thousand dollars to create the categories, but I could have done so by as little as, say, five thousand dollars to as much as 500 thousand dollars (and I would have ended with a different number of possible answers). If we need the actual dollar amount of the income reported by each respondent, we’ll chose to ask Q3.
The way we choose to create categories or not depends on the type of answers that will be suitable for our study and what type of information we want. As well, Q2 offers only two possible answers, yes or no. If the relevant information for our study is whether household annual income is below or above $50,000 (say, because the average such income has already been established as $50,000), Q2 would be the way to go.
Keep in mind that how a variable is operationalized depends not only on the researcher’s goals and needs (and practical considerations like time and money) — but also on their personal beliefs and preferences, the time period in which they live/d and work/ed, etc. Operationalizing concepts considered controversial at a specific time and place can be quite political and itself become a controversy. Consider the following example.
Example 1.2. Operationalzing Gender
It should come as not surprise to anyone studying sociology that how people operationalize gender has changed over time. Until recently, the conventional operationalization went something like this:
Q1. Are you…?
a) Male
b) Female
With advances in the study of gender and sexuality, over time our understanding of gender changed. Nowadays you are far more likely to see an operationalization similar to the following style of the American Sociological Association when collecting information on their members:
Q2. What is your gender? Select up to two.
a) Female
b) Male
c) Transgender female/Transgender woman
d) Transgender male/Transgender man
e) Gender queer/Gender non-conforming
f) Different identity (please specify) ……
g) Prefer not to state
In countries like Canada, using Q1 nowadays would might be considered too restrictive for many purposes, and also offensive by some. On the other hand, in some countries (like in Eastern Europe) choosing to go with Q2 might be seen as quite controversial and as political activism. Even in Canada the switch to more inclusive gender oprationalization is gradual and quite recent. As you will see later in the book, datasets collected in the past typically use a binary operationalization of gender.
Before we continue with measurement in the next section, here is a practical tip when working with SPSS.
SPSS Tip 1.1. Exploring How Variables in a Dataset Have Been Operationalized
When exploring an existing dataset in SPSS (more on that in Chapter 2), you can see a variable’s categories/values in the Values column in Data View. (You can switch between Data View and Variable View by clicking on their respective tabs at the bottom of your primary data window.) Clicking on a variable’s cell in the Values column will open a new window listing all the categories/values through which the variable has been operationalized.
- There are various ways one can measure concepts. At the most fundamental level, this depends on what the chosen method of inquiry (or, research) is, qualitative or quantitative. We shouldn't reify the boundary between quantitative and qualitatve methods, however. Many scientists mix their methods, employing both methods in a single study with considerable success. Social scientists use statistics predominantly when they have chosen a quantitative method of collecting and analyzing data, so here we'll focus on the quantitative operationalization of concepts. ↵