Levels of Measurement

How things are measured is of great importance, because the method used for measuring the qualities of a variable gives researchers information about how one should be interpreting those measurements. Similarly, the precision or accuracy of the measurement used can lead to differing outcomes of research findings, and it could potentially limit the statistical analyses that could be performed on the data collected.

Measurement is generally described as the assignment of numbers or labels to qualities of a variable or outcome by following a set of rules. There are a few important items to note in this definition. First, measurement is described as an assignment because the researcher decides what values to assign to each quality. For instance, on a football team, the coach might assign each team member a number. The actual number assigned does not necessarily have any significance, as player #12 could just have easily been assigned #20 instead. The important point is that each player was assigned a number. Second, it is also important to notice that the number or label is assigned to a quality of the variable or outcome. Each thing that is measured generally measures only one aspect of that variable. So one could measure an individual's weight, height, intelligence, or shoe size, and one would discover potentially important information about an aspect of that individual. However, just knowing a person's shoe size does not tell everything there is to know about that individual. Only one piece of the puzzle is known. Finally, it is important to note that the numbers or labels are not assigned willy-nilly but rather according to a set of rules. Following these rules keeps the assignments constant, and it allows other researchers to feel confident that their variables are measured using a similar scale to other researchers, which makes the measurements of the same qualities of variables comparable.

These scales (or levels) of measurement were first introduced by Stanley Stevens in 1946. As a psychologist who had been debating with other scientists and mathematicians on the subject of measurement, he proposed what is referred to today as the levels of measurement to bring all interested parties to an agreement. Stevens wanted researchers to recognize that different varieties of measurement exist and that types of measurement fall into four proposed classes. He selected the four levels through determining what was required to measure each level as well as what statistical processes could reasonably be performed with variables measured at those levels. Although much debate has ensued on the acceptable statistical processes (which are explored later), the four levels of measurement have essentially remained the same since their proposal so many years ago.