By far the most popular scale asks respondents to rate their agreement with the survey questions or statements, and we will use this scale in our examples. After you decide what you want respondents to rate (i.e., liking, agreement, etc.), you need to decide how many levels of rating you want them to be able to make. In other words, how fine a distinction do you want to be able to make between those who agree, or between those who disagree? Decades of psychological research has shown that a 6-point scale with three levels of agreement and three levels of disagreement works best. The resulting scale is:
This scale affords you ample flexibility for data analysis. Depending on the questions, other scales may be appropriate, but the important thing to remember is that it must be balanced, or you will build in a biasing factor. For example, a recent poll of the U.S. Post Office conducted by a “household name” survey company included the following scale:
ExcellentVery GoodGoodFairPoorDon’t Know
The data will be artificially skewed to the positive with this scale because there are far more positive than negative rating options. It is not just the language of the survey questions, then, that can skew your data, making scores either more positive or negative than they really are. The scale itself can skew your data, as well. Also, the statistical analyses must be appropriate for your study and your body of data or, however sophisticated and impressive, the numbers generated that look real will actually be false and misleading without warning. The validity of your survey results must be based upon scientific research principles.
Surveys created by trained researchers will not contain a ‘neutral’ or ‘I don’t know’ point in the scale. Including a neutral point negatively impacts your data on many different levels, which can have a huge, negative impact on your survey results, and there is a more immediate financial reason for not doing so, as well. We know that people harbor opinions about virtually everything, including things on which they think they have no opinion. The President of an East Coast survey consulting firm suggested that we consider “vanilla ice cream”, as an example, in defense of their use of a neutral point. He expected us to be neutral about it. But we were not. We said we liked vanilla ice cream. When people are encouraged to make a selection, one way or the other, they are able to do so, whether it is like/dislike or agree/disagree, even if they lean that way only “slightly”. Providing them with a “no brain” option only allows them an escape route from having to think, and costs you data. With a 5-point scale (two positive, two negative, and one neutral), each point represents 20% of your data. If one of those points is “I don’t know”, you could throw away up to 20% of your total survey dollars on no information at all. It is best to direct respondents in the instructions to skip questions that don’t pertain to them, and better still to not have irrelevant questions on the survey in the first place! Questions for special groups of respondents should be placed together at the end of the survey with clear instructions as to whom the questions are for. Then, we can delete any responses from respondents who answer those questions from inappropriate demographic codes.