The goal is to describe or characterize the values of a single variable. Note: to make inferences about the values of a single variable, use "Difference" below.
Differences asks whether there is a difference between one group and an expected, a priori, or hypothesized value for that group or whether there is a difference between two or more groups. Examples: Does the body temperatureIs there a difference between the scores in the control and treatment groups? Are males taller than females? Are there resulting differences in gas mileage for three different fuel additives?
Relationship between variables asks whether one variable goes up as another variable goes up or down. Thus, a relationship question requires data from the same cases on two or more variables. Examples: Is there a relationship between a student's SAT score and his or her first-year GPA? Is a car's weight a useful predictor of its gas mileage? Are people's cholesterol levels related to their age, amount of weekly excercise, and how many alcoholic drinks per week?
Caution: For some datasets, it is possible to ask both kinds of questions. Check to be sure which type is implied by the substantive question. Example: Performance is measured both before and after a training program. One could ask the difference question--are scores different after training than before? One could also ask the relationship question--are scores after training predicted by or related to scores before training.
© 2002, Gary McClelland