Two kinds of tests

Topic: Validity and reliability - issues and discussion Date: 4 December 2006 1:45 PM
Subject: Two kinds of tests Author: Mondy, Steven View

Trying to clarify some ideas, I wish to talk about the following case study, that I witnessed.
Case Study:
A teacher for some reason gives students a surprise test (with no previous warning) and then chooses to give them the same test again, but this time with warning.
The student’s are let in on the process of choosing an alternative testing date.
Alternative number 1: One week later (teacher’ choice)
Alternative number 2: Two weeks later (student’s choice)

Surprise tests are not really fair, in that they take students off guard, putting them into a situation in which they are highly resistant to the process at hand and encourage negative attitudes to the test, subject and teacher, which promotes terrible backwash. However, in such a case as above, where the teacher is trying to gauge student’s ability in a spontaneous situation and correlate it with the students prepared-for responses, then I imagine that alternative number 1 would be the better choice to maintain reliability, but would it be invalid, as the students may feel that they need more time to prepare for the test? However, is it really possible to correlate the two tests anyway? Can the formula for working out the correlation between tests be adapted to this scenario (Taking a student’s impromptu ability and comparing it to their prepared answer, I mean)? I think that it would be quite a stretch, without some sort of criterion referencing, and matching to the course objectives.
If alternative 2 is chosen, then there is too much time between the administration of one test and that of the other ~ and the test would become unreliable due to things such as ‘Test Wiseness, ‘ among other factors. (Study book ~ module 4: 4.8)…
The teacher in the above example seems to be in a no-win situation all round. The whole situation seems to be flawed anyway. The teacher will need to consider the reason for giving this test and what information is actually retrieved, and how beneficial it is to the whole learning environment. At the very least, if the tests were to be undertaken, the teacher will need to warn students of the likelihood that this kind of testing will be given.
I wonder, in what situations these kinds of surprise tests could be used in the classroom? There could be a situation, in some speaking classes, where there is a need to know how the student reacts to a new situation, based on their working knowledge of the language. Unfortunately, when in the real world, we cannot prepare for all situations. So, why not have tests that mimic this communicative task?
Steven Mondy