A topic in research methodology
"randomness is a type of lawfulness rather than lawlessness"
Mario Bunge
Randomisation is sometimes an essential technique that must be employed to draw inferences about the statistical significance of results:
"the null hypothesis, can be refuted only if the groups have been built at random."
Bunge, 1967/1988
Criterion for randomisation
"If we select students at random, then all students have an equal chance of being in the sample, and it becomes very unlikely we would get the same sample if we repeated the process."
(Taber, 2013: 255)
Can we check the process is random?
As an example consider you needed to divide a sample of twenty teachers into two groups by random assignment, and you had an alphabetical list of names.
Consider the following possible procedures:
- Divide the list of names into the top group ten names, and the bottom ten names.
- Put each name on a separate slip of paper, fold them over, and put them in a box. Shake the box, then take out ten names for teachers in one group. (The other names become the other group.)
- Go down the list of names tossing/flipping a coin for each name, and list separately those where the coin lands 'heads-up' and those who are 'tails-up'. Stop when there are ten names in one list. The remaining name(s) on the original list go(es) onto the other group list.
- Number the names in the list. Put odd numbered names into one group, and even numbered names into the other group.
- Go down the list of names rolling a die for each name, and list separately those where the die lands with an even number (2, 4, 6) or with an odd number (1, 3, 5). Stop when there are ten names in one group list as the remaining names go in the other group list.
Ask yourself what would happen if you repeated the process several times. If the process would give the same groupings each time (as it would in two of the suggested procedures), then this is not a random process.
It might be argued that in the other cases the process is not really random as if the conditions were precisely repeated (e.g., tossing an identical coin in exactly the same way, from the same height, in precisely the same place, with just the same amount of force, in air of the same temperature and wind/breeze conditions…) then the outcome would be the same. Even if this is, in principle, true, such replication of conditions is not possible in practice, so this would be considered 'good enough'.
(Read about 'Nothing random about a proper scientific evaluation?')
Sources cited:
- Bunge, M. (1998). Philosophy of Science. Volume 1: From problem to theory; Volume 2: From explanation to justification (Revised ed.). Routledge. (1967)
- Taber, K. S. (2013). Classroom-based Research and Evidence-based Practice: An introduction (2nd ed.). London: Sage.
My introduction to educational research:
Taber, K. S. (2013). Classroom-based Research and Evidence-based Practice: An introduction (2nd ed.). London: Sage.
