Levels of MeasurementLevels of measurement concern the precision of data you gather about characteristics, or variables, of interest in a study. Measurement scales allow you to convert observations about variables into numbers. The precision of these numbers influences which statistics are appropriate and how they are interpreted.For example, seating preference is a variable of interest to the airline industry. Options that travelers may select could include: (1) Window, (2) Center, (3) Aisle, and (4) No Preference. Numbers that are assigned to each option for the data analysis have a great deal to do with how results are interpreted. 
The nominal scale is the least precise. Nominal data is derived from qualitative categories--names, such as those in the airline example. Gender, religious preference, and yes-no responses are other examples. Numbers assigned simply categorize for analysis. One is not better than another.
The ordinal scale adds a level of precision. It measures the relative position of objects or individuals, but does not indicate the distance between them. Examples of ordinal data include class ranking and finishing order in a cross country race (e.g.) first, second, third). Here, the ranking does indicate placement, but does not show how far apart one position is from the next. The interval scale is even more precise. With interval data, distances between positions are equal, but "0" is an arbitrary assignment. For example, with temperature, each degree is equally distant from another, but "0" does not mean that there is no temperature. It is simply a reference point on the scale. The ratio scale is the most precise. With ratio data, all positions are equally distant and "0" means that the value is truly "0". If you have "0" money, you have none. But if you have $200, you have twice as much as a friend who has $100. One of the assumptions for statistical analyses concerns the levels of measurement. For example, you would have a challenge interpreting the mean (arithmetic average) for religious preference, since it is categorical. But you could use the mode as the average to show which one was selected most frequently. The more powerful parametric statistics (e.g., t-test, ANOVA) are applied under certain assumptions, one being that data are at least interval; hence, the results are more robust.
|
