Dr. Dennis Gleiber Explains How Limits Can Impact Satisfaction Scores
Everyone wants to get high scores. We learned this all through life, for example in school and from sports. We go on and generalize this orientation to most everything else in life including business. High scores are GOOD! Of course high scores or high values are indeed good especially when the referent is without upper bound. More money is good; more sales are good. When the referent and its measure are unbounded, like sales or sports scores, higher is better because higher is always possible and highest wins, thus ostensively, best.
Is a 4.0 as Good as it Gets?
When the referent and its measurement are bounded, extreme values like high and low are no longer the same as when unbounded. For the unbounded, “the sky is the limit,” but for the bounded, there are limits. Thus we quip, “I was so low the bottom looked like up”. Consider grades in school. “A” is an upper limit. You can do no better, but it does not matter that much because what you are graded on is also bounded, the one test or the one course or the one matriculation. Achieving or obtaining the highest score has no direct impact on the next measurement. [This is an interesting case because empirically higher grades are a good predictor of higher future grades and at the same time regression to the mean is still in play and lower grades may be forth coming. An overall GPA of 4.0 is not that uncommon but likely results from the arbitrary limit on the upper bound of grading at “A’.]
Revenues and profits are unbounded in either sense but other measures that are important for measuring success in business are bounded by convention. Satisfaction and other evaluations are bounded by the method of measurement, which in survey research is the scale of the response options. The satisfaction score can be no higher or lower than the responses allow. It is also limited in variability based on the number of options provided. The former means that there are limits to how high the satisfaction score can go. Even if the end points refer to an ideal like “completely” satisfied or dissatisfied which are theoretically true limits, the fact that they are offered means that they will be selected by at least some respondents. [Notice that someone who responds completely satisfied very likely be even more satisfied at the next measurement although regression to the mean or comparative evaluation suggests they will not be.] The latter means that although the measurement is conceivably continuous over its range, in practice it is not and therefore is more accurately described as ordered categorical. Thus, there are limits on variability within the distribution of the measure as well as at the extremes of the measurement.
Why Does All of This Matter to Business Decision Makers?
The simple answer is that measures like satisfaction are not conceptually consistent with how we understand scores. While higher may still be better, higher no longer means the same thing and in the limit is not even possible. Business decision makers who are interested in satisfaction often want a certain level of satisfaction. However the higher the bar is set the more difficult it becomes to reach the standard. In fact, when the standard reaches a relatively high level it becomes untenable for several reasons. The first is psychological. Remember, with satisfaction we are thinking about how human beings evaluate. They must compare their reaction to the external stimulus of the product or service, an internal response, and then compare that orientation to the range of external response options. This is archetypal behavioral psychology in which we are trying to measure the internal orientation based on external but contrived behavior (survey questions). Notice the immediate disconnect between reported satisfaction and observable behaviors of interest like purchase, retention, etc. These two stimulus response mechanisms are probably neither fixed, nor consistent. The volitional elements of the two stimulus-response reactions are likely to be changing based on changes in expectations about the external stimulus itself and the context of the experiences with it.
Dennis Gleiber, Ph.D., Chief Research Scientist
Next week, Dr. Gleiber will explore how rising scores may distort our views of satisfaction.