Extraneous variables are a challenge to both the internal and external validity of the experiment. Against the willingness of the research and the researcher, they tend to have an impact on the dependant variable and affect the outcome of the experiment. In the process of research, there is a need to control the extraneous variables as they add an alternative explanation of the results.
Largely, there are four approaches by which the effect of the extraneous variables can be controlled.
1) Randomization: In this approach, treatments are randomly assigned to the experimental groups. It is assumed that the extraneous factors are present equally in all the groups. This technique is only workable when the sample size is very large.
2)Matching: Another important technique is to match the different groups of confounding variables. Different confounding variables like gender, age, income etc. could be distributed equally amongst the group. It sometimes does become difficult to extend matching within all the groups and another drawback of the same is that, sometime the matched characteristics may be irrelevant to the dependant variable.
3)The use experimental designs: In certain studies, the experimental designs may play a crucial role in reducing or completely removing the role and impact of the extraneous variables.
4)Statistical Control: There may be situations, when all the above mentioned methods to control the extraneous variables do not show any significant outcome. It brings the entire research into question as then causal inferences are difficult to make. Another method that may work to bring down the effect of extraneous variables is the method of statistical control. Among the various statistical tools and techniques, Analysis of Covariance ( ANOVA) helps in reducing the impact of the extraneous factors on the study.
These four methods, in their own way, can be used in the research, collectively or exclusively to eliminate the relationship impact discussed above. It is dependent upon the expertise of the researcher to understand and administer these methods in a way that the best possible results can be obtained.