Proper scale labeling improves the chance of accurate, helpful results. One element that makes scale labeling challenging is learning the difference between bipolar and unipolar scales.
Here is an example of a bipolar scale:
This scale works because cold is the opposite of hot, and there is a neutral point, since something can legitimately be neither hot nor cold.
Here is an example of a unipolar scale:
This same unipolar scale could also be constructed as:
Notice there is no neutral point on the unipolar scale. We could potentially create a midpoint of “half full and half empty,” but inexperienced authors will often try to create that label as “neither full nor empty.” This simply does not work. Empty is NOT the opposite of full in the same way that cold is the opposite of hot. Empty is just another way of describing the “degree of full.” “Neither full nor empty” cannot be the mid-point of such a scale.
While the full/empty scale may seem obvious, there are many situations where inexperienced authors fail to recognize the difference between unipolar and bipolar scales. One frequent problem area is importance scales.
Here is an established reliable importance scale:
There is no mid-point to this scale. Frequently, in an attempt to create an importance scale with a mid-point, we see people create scales such as:
This has proven NOT to be a reliable scale. “Neither important nor unimportant” cannot be the mid-point of this scale as it is the equivalent of neither full nor empty.
Once an inexperienced author has given up on labeling a scale, the number of scale points chosen can exacerbate the issue. The number of points on a scale should be a tradeoff between sensitivity and reliability. For example, suppose we want respondents to rate McDonald’s Big Mac. Consider the two scales below:
The first scale would be very reliable. However, it would not be very sensitive. The scale only has two categories – good and bad. This means it would not be sensitive enough to distinguish between how good and how bad.
The second scale has been established as reliable. However, inexperienced authors will often try to extrapolate – by thinking, “If the 5-point scale is more sensitive than the 2-point scale, wouldn’t a 100-point scale be even more sensitive?” Well yes, the 100-point scale would be more sensitive. However, unlabeled, it would not be very reliable – and good luck labeling a 100-point scale. It’s crucial that authors find the perfect balance between sensitivity, reliability, and practicality when creating scales.