Scales are often used in questionnaires to measure variables like "level of satisfaction". Even in the simplest case, it is important to be careful in order to avoid any mistake. One cause of problem, can be the past experience with questionnaires of those you plan to interview. Take the example of a simple scale (with 5 levels). Then, imagine that these levels are the following:

Then, imagine a questionnaire that use that kind of scale. Each person who will answer must then, encircle the number which corresponds to its answer. Who can be that person ? A customer that just entered the store! Before going further, keep in mind that this person has, most probably, others concerns than our survey. What will happened ? He might show some goodwill, and he will answer the survey. But, he might answer according to the kind of instructions he use to encounter in others questionnaires he answered . Thus, even the clearest instructions might be ignored if our "customer" is accustomed to other instructions. Let go back to our example of questionnaire, to discover how it is potentially flawed.

First of all, there is the choice of numbers. By using 1, 2, 3, 4, 5, (in this order) we are choosing the concept of rank. In this logic, the first position is always the best. But, everyone who went through some school exams, know that many people are also accustomed to see pointing. So, only 1 point represents the lowest value. It is the opposite of what is said by the descriptions in our little scale.

To avoid that problem, we can repeat our choice of answers at each question. The idea seems interesting, but it might be a trap. Let us not forget this : each time a survey became longer, our rate of answers is likely to decrease. The size of a survey, is not only the number of questions. It is also its physical size, in other words : the size of the text! Thus, the repetition of the choice of answers must be used with caution. A wise solution would be to reformulate our questions in terms of pointing. This last solution refers to learned behaviors then, it might has less of chances to be wrongly understood.

There is still another problem. The choice of the descriptive terms. By using the expression "entirely satisfied", we can designate a few kind of situations. Here are two possibilities:

Several people will tend to interpret the "first position" of a scale like it should mean something special. We are in a society where the concept of competition is etched in many facets of our lives. Thus, the first place of something must be special. What reinforces this point of view, is that undecided people have a tendency to choose the answer in the middle of a scale. On the another hand, the terms used in our scale could means that if the customer is simply "satisfied", there is a problem somewhere. Let us look at what occurs in the real world:

Let us imagine that our customer considers that he had an adequate service. If he understands that "entirely satisfied" means "service with no problem", he will choose "1". But, if our customer understands that "entirely satisfied" means "outstanding service", he will rather choose 2 or 3. As you can see, our simple scale can be understood in many ways.

The lesson of this story? One is likely to obtain unusable information if he do not care about the past experience with questionnaires within the group he plan to interview.

And in our example, what can be the solution ? It will be necessary to think about the precise objective of the survey. Does one wish to know if there is there a problem ? What is the amplitude of that problem ? If it is the case, the good service will deserve a perfect score. It is then necessary to take care of how the question is asked. Or, one seeks to identify a level of satisfaction, in which case, the good service, will deserve a "medium" score. In such case, the formulation might be less tricky.