go to the School of Psychology home page Go to the UNE home page
Chapter 7 - Analysing the Data Part IV - Analysis of Variance Chapter 1 - Behavioural Science and research Chapter 2 - Research Design Chapter 3 - Collecting the Data Chapter 4 - Analysing the Data Part I - Descriptive Statistics Chapter 5 - Analysing the Data Part II - Inferential Statistics Chapter 6 - Analysing the Data Part III - Common Statistical Tests Operationalism Experimental and non-experimental designs Internal and external validity Between groups Vs repeated measures designs Ethical issues


Chapter 2: Research Design


Non-experimental designs

Perhaps the simplest design is the correlational design or quasi-experimental design. A study qualifies as correlational if the data lend themselves only to interpretations about the degree to which certain things tend to co-occur or are related to each other. For example, a social psychologist might be interested in the degree to which children who watch violent television shows tend to aggress at their classmates on the playground. To conduct this study, a set of children are observed for a period of one month by a set of trained observers, who record the number of violent acts the child engages in (e.g., hitting another child). Each childŐs parent is asked to answer a set of questions about which television shows the child has been exposed to in the last month. Of interest in this part of the data collection phase is how many, and how often, certain television shows watched by the child in the last month were classified by the Australian rating system as containing violence. At the end of the data collection, the research would have recorded for each child a number reflecting his or her violence on the playground, and a number reflecting the amount of violence the child is exposed to on television. Of interest to the investigator is whether violent behaviour (high "scores" on the measure of playground violence) tends to occur more so in children with high "scores" on the measure of exposure to television violence (and vice-versa with low "scores" being paired with low scores). If so, we can say that exposure to violent television is correlated positively with actual violent behaviour.

The major advantage of correlational designs is that they are typically fairly easy to conduct. The major disadvantage of correlational designs is that they leave the actual reason for the associations found quite unclear. Suppose, for example, that there was a positive correlation between exposure to violent television and violence on the playground. You might be tempted to conclude that exposure to violent television causes children to be aggressive and violent. But such a conclusion is completely unwarranted. Rarely does a correlational study allow inferences about cause and effect. In this case, there are many other possible explanations for the relationship. For example, perhaps children with parents who neglect and physically abuse them just plop them in front of the TV at night, when the violent shows tend to be on. These children may have learned from the parentsŐ abuse and neglect that aggression and violence are acceptable ways of relating to others, and so they do it on their own around their classmates. In such a case, the obtained association between exposure to violent television and actual violent behaviour may be spurious. That is, a third variable Ń parental neglect and abuse cause them both. Another possible explanation is that children who tend to be violent, for whatever reason, tend to prefer watching violent television shows. It isnŐt that the violent television causes the violent behaviour. Instead, some children are dispositionally violent (that is, due to personality or genetics Đ also Ôthird variablesŐ) and it is this that determines both the preference for violent TV shows and the playground aggressiveness.

In spite of these problems, correlational studies are quite common and popular. This is probably because the discovery of association suggests the possibility of cause (the second bullet point given in Chapter 1). That is, while the fact that two things are correlated doesnŐt allow us to directly infer causation, if the two things are causally related, they must be correlated. So showing correlation can be a useful first step toward demonstrating causation.




© Copyright 2000 University of New England, Armidale, NSW, 2351. All rights reserved

UNE homepage Maintained by Dr Ian Price
Email: iprice@turing.une.edu.au