Education Research Impracticalities

Pure research is aimed at having something to say. Practical research is aimed at knowing what to do. It leads to action.

The hi-tech industry’s research labs generate the type of practical knowledge that can be translated into hardware and software: For example, IBM’s labs kept generating knowledge that enabled IBM to produce a faster, more reliable, less expensive computer hard-disks. Similarly, pharmacological research labs generate the practical knowledge necessary to produce medications with higher impact, lower side-effects, and longer shelf-life.

In social sciences in general, and in education in particular, research is something else. One of its aims is to generate a better understanding of how people behave, learn, etc. Often, though, we settle for just feeling better about our understanding, and don’t necessarily generate real knowledge: The type that can makes predictions that are verifiable and refutable. The type on which we can base action with some confidence. What if we try to cause a larger portion of education research to be of a nature similar to the research described above, used in different disciplines of engineering? We might get more actionable knowledge which will enable us to create educational environments and methods that will lead to better results, in a somewhat predictable manner. What a concept.

One main difference between engineering research and social-science research is that much of social-science research is qualitative, while in exact sciences and engineering, any research that is not quantitative wouldn't be considered research at all.

There are many good reasons for the use of qualitative research in education, and for avoiding much quantitative research:

One reason is that to measure anything, and to make a clear statement about the measurement, one requirement is the isolation of a single variable, or a very small set of variables. In social sciences such as education, this is prohibitively difficult. For example, trying to research a common question of interest to many: What effect does class-size have on the quality of education? There are many aspects of “quality of education”, so we would have to concentrate on a subset so small that it may be meaningless. A tempting, maybe even reasonable, focus may be the grades pupils receive. A quantitative research may analyze the statistical correlation between class size and grades. But how do we make sure other facts (variables) don’t get in the way? Whole classes of variables such as socioeconomic background, quality of teachers, the school infrastructure, pupils’ expectations of themselves and of the school, the social environment outside school, other schools in the vicinity, the characteristics – such as size – of the classes in previous years, the temperature, lighting and air quality in the class, the amount of ambient noise inside and outside the class, etc. It is truly difficult to isolate the interesting variables from all the others, when conducting research in the field.

A controlled lab experiment can do a much better job in isolating variables, but it has its own problems: Putting a small number of children in a lab for a short time experiment will not tell us much. Also, it is quite difficult to get around some legal and ethical issues, and your garden variety doctoral student doesn't want to concentrate on that. If we try to put a meaningful number of pupils (how about 2000?) in a lab, we run into major logistical problems. Then there is the timeframe issue: To make a meaningful statement, we may want to run an experiment for days, months or years. This exacerbates the logistic and ethical problems to a point that makes the notion ridiculous.

One might try to escape to just sending questionnaires and having people – pupils and former-pupils – answer them. Then surely it would be possible to hammer the raw questionnaires with statistical tools and get useful knowledge. Except one would have to first formally validate the questionnaires. This turns out to be such a significant undertaking, requiring expert attention, time, manpower and money, that “one” tends to be rather discouraged from doing it.

Ok, then. How about collecting masses of raw data about people, schools, etc, and analyzing that? Pupils' grades over time; pupils attendance information, pupils families' socioeconomic data, graduates' employment data, etc. Good luck getting the raw data. There are legal problems, privacy issues (some real and some used as a convenient obstruction technique), and the fact that the data is often not collected and stored systematically. It is next to impossible to obtain significant amounts of continuous data. Very few researches succeed in doing that.

Then there is the difficulty and academic disinterest in conducting follow-up research. There is the same problem with research aiming to verify or refute a previous research. There are potential political problems with generating knowledge about issues with social significance: The knowledge may contradict someone's beliefs, and we don't like our beliefs to be challenged. These days it is not such a big problem when researching physics, for example, it used to be - just asks medieval scientists. But it is always a problem with social sciences, which directly relate to our political, moral and religious views.

So, education researchers often fall back on the relatively safe field of qualitative research. And instead of using qualitative research as a starting point that leads to a more quantitative, accountable and actionable research, we are left with this qualitative exploration as the final product.

Assuming that we want to encourage more educational research that will conform to similar guidelines as those used in exact sciences, we would need to deal with many of the above issues. We don't need to make the problems go away completely; it's enough if we make it easier for researchers.