In psychology we have a saying, “Correlation does not equal causation”. Correlations exist between pairs of things that are systematically related. To understand cause and effect, we need to conduct an experiment. Understanding whether something is correlational or experimental can be challenging at first.
When results are correlational, generally the researchers obtained two “scores” – sometimes two from the same person (such as height and weight), and sometimes two from 2 different people such as husband’s height and wife’s height. They would obtain the scores from a sample – a large group of people. Everyone in the study provides the same scores, and the correlations are calculated for pairs of scores.
In an experiment, look for different groups of people doing different things. There is always an experimental group (also called the experimental condition) and this group is “testing” the variable that the experimenter thinks will cause an effect. In a drug study, this is the group who gets the medication. The experimental group has to be compared against a control group – this group does NOT take the medication in a drug study – they get a placebo instead. In an experiment the experimental condition scores will be compared to the control group scores.
One final word about correlations. The examples we have been looking at this week are concerned with two variables that show a linear relationship – for example, as scores on one variable go one way (e.g., hours of studying), scores on another variable go either the same way (e.g., GPA) or the other way (e.g., errors on exam questions).
There are relationships between variables that are non-linear. Consider the classic upside-down U-shaped relationship between physiological arousal and performance, showing that a medium level of arousal is best for performance, while extremes in either direction (total boredom/apathy in one direction and panic in the other) impair the ability to get the job done:
