In an experiment, the researcher manipulates the independent variable(IV) to see if a change can be induced in the dependent variable(DV) while confounding variables are sometimes controlled. Experiments aim to test the relationship between variables, which can be either:
- Correlations: two or more things occur at roughly the same time. This suggests a possible relationship that is not necessarily causal.
- Causation: when A happens, B must follow as B is caused by A. This suggests a relationship of causal and predictive nature.
Lab experiments takes place in a laboratory, where conditions can be monitored and controlled. An example of a lab experiment is Bandura et al.'s (1963) 'Bobo doll experiment'. It showed how violent behaviour represented in the media(IV) can affect children's behaviour(DV). Three experimental groups and a control group were used.
- Experimental group: subjects of the experiment. IV is manipulated (in this case, films depicting violent behaviours were shown to the children)
- Control group: another group with matched characteristics of the experimental group, however, the IV will not be manipulated (in this case, the children were not shown violent film)
Bandura et al. found at worst a correlation and, at best, a causal relationship between seeing violence and acting violently. To further separate correlation from causality, the test-retest method can be used. If replicating the experiment (with different participants, for example) constantly achieve the same result, the relationship established is more likely to be causal.
- Easy to replicate
- Can establish cause-and-effect relationships
- Hawthorne/observer effect
- Artificial environment reduces ecological validity
Field/natural experiments take place in a natural setting, which is much more commonly used than lab experiments in Sociology. Since it's more difficult to exert control over the environment/conditions, correlations rather than causations are more likely to be established.
Rosenthal and Jacobson(1968) conducted a field experiment inside a school on how expectations from the teachers about the ability of their pupils(IV) can affect the educational achievement of the pupils(DV). They gave a sophisticated IQ test to the pupils, randomly selected some students and then told the teachers that those students would display 'dramatic intellectual growth'. They retested the pupils at a later date and found a significant increase in the IQ scores of the 'late-developers'.
- High ecological validity
- Participants behave more naturally if unaware of being studied
- Less likely to control the conditions
- Unethical to study participants (if) without consent