EDEV_507 Week 8_6

Here’s my take on the variable/category issue. As usual, much of the confusion arises from similar terminology being used in different ways.

In virtually all stats textbooks, the term quantitative ‘variable’ is explained as something that can ‘vary’ and that is of interest in a research design. Variables fall into a number of types: nominal, that is, those whose name (from the Latin ‘nomen/name’) defines the variable, such as ‘gender’ (leaving LGBTIQ aside for now) , ‘university-educated or not’ and other binary or small categories of information. In these cases, there is no gradation between categories. Someone is either ‘university-educated’ or they are not. These are also often called ‘categorical’ variables, but this use of ‘category’ is not the same as what José referred to; ordinal, that is, variables that can be ‘ordered’ in a line (from the Latin ‘ordo/line or sequence’). Ordinal variables have a hierarchy. Some are numerically bigger than others, but the distance between the variables is not known. A Likert scale, with options 1 to 5, is this type. There is no way of knowing how different ‘1 to 2’ is from ‘4 to 5’.  Understanding how the use of ordinal variables limits research findings is an important task. (For example, the median statistic is predicated on the distance between the Likert options being the same. But if they are not, in reality, the results will be compromised.); continuous, that is, a variable that is ordered—like an ordinal variable—and whose distance is known. My height is different from you, and the degree of difference can be known as much as the preciseness of the measuring instrument allows. Arguably, it will be recognised that only continuous variables can exist, but for ease of research, the others are used routinely.

The fundamental feature of quantitative research is the positivistic belief that human activity can be measured and translated into continuous variables (and the others, but let’s leave that aside here). However, two major problems have to be dealt with. The first is understanding of, and access to, hidden activity, and the second is that that activity often has no corporeal form. Can you measure my happiness? You might decide to define happiness as the width of a smile, or the increased heartbeat, or the particular neurological change in the brain, or with some other measurement. No one can feel another’s happiness. The existence of happiness can never be known directly and this kind of notion is called a construct. Of course, the selection of any proxy measurement for happiness must contain assumptions. I can smile widely when I’m angry (to hide my distain); my heartbeat will be faster during running away from a bull. In order to make the measuring instrument more precise, a single invisible activity may be defined as the result of a group of measurements: the width of the smile, the degree of heartbeat change and the neurological movements. Noting how variables change together is the basis of (multi-) variate analysis, and combining variables into a single package is the basis of factor analysis.

In qualitative research, the purpose and function of variables doesn’t seem (I say seem because I’ve never actually done any qualitative research before) different. The processes and techniques are entirely distinct, though. Let’s say I want to measure the degree of happiness in a participant. Well, I could just ask them, but I wouldn’t know how happy they were. Well, I could just ask them again, this time adding a qualification statement like, ‘Compare your current happiness to the happiest time you’ve ever been. How different is it?’ But then, I wouldn’t know how happy Participant 1 is from Participant 2. How could I tell the difference? Maybe the exact difference is not important, only that there is a difference. However, in all cases, I need to rely on participants telling me the truth. I need to have tools and techniques for ascertaining the veracity of participant information. Notice that the questions asked just got longer until the degree of happiness was found.

This style of questioning presumes the existence of a hidden construct of happiness. But some qualitative research methods (grounded theory, for instance) does not begin by accepting the existence of the construct of happiness. Rather, information is collected on a more general theme, and through interpretation (coding, memos, constant comparison), the existence of hidden (well, constructs are always hidden) constructs becomes a possibility. The task then is to selectively limit the data collection until the construct is either shown or not. ‘Construct’ is what José means by ‘qualitative variable’ as it relates to qualitative research, although he used the terms ‘category’ and ‘theme’. Fundamentally, they are the same, and in many research designs, quantitative confirmatory research often follows qualitative exploratory research on this basis. But now we are getting on to mixed methods, and that’s for next week.


About theCaledonian

Scot living in north Japan teaching at a national university.
This entry was posted in EDEV_507. Bookmark the permalink.

Leave a Reply

Please log in using one of these methods to post your comment:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s