Intuition can take you a long way in business, but there are times when you need hard data to analyze and numbers to crunch. The Six Sigma process for achieving top quality business processes defines several different types of data. Attribute data is of the yes-or-no variety, such as whether a light switch is turned on or off. Variable data is about measurement, such as the changing light levels as you adjust a dimmer. They're both important information, but variable data is usually more useful.
Attribute data focuses on numbers, variable data focuses on measurements. For example, suppose you're gathering data on defective products that your assembly line turns out. Attribute data simply classifies the output as defective or not defective. If you gather variable data, you might look at how bad each defective product is: 10 percent faulty, 20 percent faulty, and so on.
Neither of these is inherently wrong. It all depends on how you want to use the data. If you're practicing the Six Sigma approach and you want to see how many products meet your high standards, attribute data might do the trick. If you want to measure the quality of each product, variable data is probably more useful.
There are other ways to classify data. Data that doesn't adapt well to numbers, such as color or taste, is called qualitative data, for instance. Attribute data is simpler to gather than qualitative data, so it's a good choice if you're looking at a binary condition, where there are just two alternatives:
- The product works or it doesn't work.
- The salesperson closed the deal or she didn't.
- The parts fit the slot they're supposed to belong in or they don't.
- Students pass the test or they fail.
You can compile the attribute data to see how well your process, equipment or staff are performing. If you want 80 percent of your students to pass their final exams, and only 20 percent do, that shows a problem. Whether it's the student body, the teachers or some other issue will have to be determined.
Variable data can tell you many things that attribute data can't. Suppose you're testing new girders for use in a construction project. Attribute data tells you the percentage of girders that bear up under the load you put on them. Variable data can tell you if a specific girder that passes the test may still be dangerously close to giving way. If you want to know how badly failing students missed passing their exams, variable data can give you the answer.
