Validity and dealing with the problems that may arise
All research strives to be high in validity. Any flaws must be minimised in order to draw valid conclusions from a study. There are two kinds of validity:
- Internal validity concerns what goes on inside a study - whether the researcher tested what he or she intended to test.
- External validity concerns things outside a study - the extent to which the results of the study can be generalised to other situations and people. The term ecological validity is often used as another term for external validity.
Internal validity may be affected by extraneous variables (EVs) that may cat as an alternative independent variable (IV). Conclusions about the effect of the IV on the dependent variable (DV) are erroneous if changes in the DV are due to EVs/confounding variables. Internal validity of an experiment may also be affected by the use of measurements that lack validity. For example, a study on gender differences in conformity may use a questionaire to assess conformity. If the questionnaire lacks validity, this means the experiment in general lacks validity. It could also be that the study assessed conformity by observing the behaviour of participants. If the observations lacked validity, this would affect the internal validity of the whole study.
External validity Many people think all lab experiments are low in external validty wheeras field experiemtns and natural experiments conducted in more natural surroundings, are seen as high in external validity. This is not necessarily true. In some cases the contrived, artificial nature of the lab setting is immaterial to ht ebahviour being observed (such as a memory task) and therefore it can be generalised to evryday situations (high external validity). In some cases, field and natural experiments can be very contrived and artificial. It is often more important to consider issues such as whether the participants were aware of being studied (which reduces the realism of their behaviour) and whether the task itself (rather than the setting) was artificial and thus low in mundane realism, which reduces the generalisability of the results.
Reliability and dealing with problems that may arise
In the context of an experiment…