Randomisation

A topic in research methodology

"a quasi-universal special method of science is random sampling, i.e., the extraction of a small subset from an original set or population (which may be infinite) in such a way that the selection does not depend on the properties of the individuals, but is blind to then and consequently unbiased."

Mario Bunge

Randomisation is a process which is essential in some kinds of research design. Experimental research that leads to a calculation of the likelihood of the outcome (to be compared with the arbitrary cut-off where a result occurring by chance is considered 'significant') often relies on randomisation:

  • to randomly assign students to treatment groups

(or less powerfully)

  • to randomly assign groups to treatments

Sometimes when we select a sample from a population it is important (to apply statistical tests correctly) we have a random sample. This means that each person in the population has the same chance of being selected for the sample. (In other kinds of research, informed by different paradigmatic commitments, we would select our sample deliberately, and not wish to rely on chance).

The use of inferential statistics to draw conclusions form experimental studies often relies upon the studies being undertaken with units (this might be students, or teacher, etc. – the 'units of analysis' used for the analysis) randomly assigned to conditions; and to generalise those conclusions from the sample to a wider population may depend upon the sample being randomly chosen from the population.

When an experiment tests a sample drawn at random from a wider population, then the findings of the experiment can be assumed to apply (on average) to the population. (From Taber, 2019)

To misapply statistics to non-randomised samples as if they are random has been considered one characteristic of pseudo-scientific work (Bunge, 1998).

How do we know a decision has been made randomly?

"Although the term 'randomly' is much used, we know that most people are not good at appreciating what 'random' really means, and expect random events to appear much more evenly spread than is often the case. … This leads to a question of how a researcher selects participants for a study in a truly random way. When we read accounts that report 'the teacher selected a student to interview at random', we might wonder how they went about that."

Taber, 2013a: 255

To make a random selection we need a procedure such a tossing a coin or drawing lots, or using a table of random numbers. In quality research, the researcher reports how the randomisation was undertaken, but often these details are missing, and sometimes the description that is given does not seem consistent with a random choice.

"This makes one wonder whether articles in the educational research literature that make an unelaborated reference to random sampling or randomisation in quasi-experimental studies always do actually refer to a valid randomisation process rather than some form of convenience sampling or tacit application of an implicit sampling frame that the investigators considered random, or 'as good as' random, at the time.

Such doubt can be avoided if when we write-up our research we always think to follow any claim about randomisation by a sentence of two explaining the technique we used to ensure that sampling or assignment to conditions was indeed random. Then readers of our work can know that 'random' means random, and not just that we asked the teacher who happened to be in the staffroom when we were getting a coffee."

Taber, 2013b: 362

(Read about 'Nothing random about a proper scientific evaluation?')

Am I really randomising?

There is a simple criterion for deciding whether a process gives random outcomes.

Example of randomisation:
  • Population – grade eleven learners (in four classes) in one school
  • Sample – two classes to be assigned as experimental and control groups

"Two grade eleven classes were picked at random…this was done by writing class names (11A, 11B, 11C, 11D) on pieces of paper, then putting them in the container and finally randomly select two papers from the container. Thereafter the classes that were randomly selected into a research sample were further randomly assigned into either experimental group or control group."

Magawa & Kalebaila, 2020

Sources cited:

My introduction to educational research:

Taber, K. S. (2013). Classroom-based Research and Evidence-based Practice: An introduction (2nd ed.). London: Sage.

facebookShare on Facebook
TwitterTweet