There are three main strategies in research, based around different ways of collecting and analysing empirical evidence:
(b) case studies; and
Experiments are not often used for the purpose of CIPD dissertations, however it might be useful to describe this strategy briefly here.
Experiments measure the effects of manipulating one variable on another variable. They involve the selection of a sample of individuals from a known population and, under controlled conditions, introducing planned change to them. The effects of that change in the sample is then measured.
The strategy is usually used for hypothesis testing and is common in many of the natural sciences where laboratory conditions can be controlled. It is, however, used extensively in the social sciences - principally to explore aspects of behaviour and personality, and in controlled statistical manipulation of figures.
The classical approach to experiments is to choose subjects at random from a known population and divide the sample into two groups - an experimental and a control group. This is often referred to as the "post-test only, two group" design. Conditions for the control group are kept constant, but those for the experimental group are changed in one respect by the researcher. Both groups are then measured in exactly the same way to determine whether there is any difference between them. Using statistical analysis, it is possible to determine the probability of any observed difference occurring purely by chance and, hence, the likelihood that the planned change was the cause. There are other, more sophisticated (and usually much more expensive) methods of testing, but this remains the simplest and most common.
However, the validity of experiments - particularly in the social sciences - is limited in two ways:
¨ internally - by the extent to which the factor or variable actually caused the effect observed (which is really a function of the validity of the controls); and
¨ externally - in respect of the degree to which the findings can be generalised from a specific sample to a population as a whole.
In addition, experimental studies often encounter problems - from both a practical and ethical point of view - when moving from the laboratory to the real world. Controlling conditions is not that easy!
Alternative approaches - sometimes referred to a "quasi-experimental" - have been developed to cope better with conditions outside the laboratory. One of the most common methods is the "pre-test/post-test" design - this seeks to establish control and internal validity by recording conditions both before and after the experiment. The same measurements are carried out on both the experimental and control group before and after the manipulation of the variable and the results compared. Any difference between the two groups, given their prior variations, may be taken as evidence of a causal relationship.