Measurement

The assessment of a trait or feature against a standard scale.

Psychologists rely heavily on measurements for very different purposes, ranging from clinical diagnoses based on test scores to the effects of an independent variable on a dependent variable in an experiment. Several different issues arise when considering measurement. One consideration is whether the measurement shows reliability and validity. Reliability refers to consistency: if the results of a test or measurement are reliable, a person should receive a similar score if tested on different occasions. Validity refers to whether the measurement will be useful for the purposes for which it is intended.

The Scholastic Assessment Test (SAT) is reasonably reliable, for example, because many students obtain nearly the same score if they take the test more than once. If the test score is valid, it should be useful for predicting how well a student will perform in college. Research suggests that the SAT is a sufficient but not perfect predictor of how well students will perform in their first year in college; thus, it shows some validity. However, a test can be reliable without being valid. If a person wanted to make a prediction about an individual's personality based on an SAT score, they would not succeed very well because the SAT is not a valid test for that purpose, even though it would still be reliable.

Another dimension of measurement involves what is called the scale of measurement. There are four different scales of measurement: nominal, ordinal, interval, and ratio. Nominal scales involve simple categorization but does not make use of the notion of comparisons like larger, bigger, and better. Ordinal scales involve ranking different elements in some dimension. Interval scales are used to assess by how much two measurements differ, and ratio scales can determine the difference between measurements and by how much. One advantage of more complex scales of measurement is that they can be applied to more sophisticated research. More complex scales also lend themselves to more useful statistical tests that give researchers more confidence in the results of their work.