Scales of Measurement and Answering Categories
by Clement LoQuantitive Measurement
In a market survey, information is collected in unit measurement so that it is possible to analyze the results. With computer software, every piece of information has to be expressed in quantitative units. For instance, in studying grocery shopping, we detect its frequency by asking:
About how many times PER WEEK do you shop for groceries?
Less than once a week
Once a week
2 - 4 times a week
5 or more times a week
It is straightforward to figure out a scale to measure shopping frequency (i.e., how often). In this question, we use four answering categories. The response would tell us very clearly about consumers' frequency of grocery shopping.
Tracking behaviour
When we want to collect information on buying intentions, customer satisfaction and service assessment, these issues require different scales of measurement. We have to be more ingenious in designing answer categories. Experience shows the better the answering categories, the better the data quality. Here are some pragmatic suggestions.
First suggestion
Keep the answering categories to a minimum. In the following example, a 10-point scale is used. I would say that a 10-point scale is the maximum.
Recalling the last time you telephoned the order desk at ABC Inc., how would you rate their service? On a ten-point scale, "10" being "excellent" and "1" being "poor," please circle the number that corresponds to your rating.
Service at ABC's order desk:
Excellent Poor
10 9 8 7 6 5 4 3 2 1
Avoid going beyond that. It is very difficult for anyone to break it down to more levels of rating. When there are many gradations on a scale, it puts great demand on respondents. You then run the risk of introducing bias (i.e., errors) into the data. Feeling overwhelmed, respondents would start to check off numbers without much thought.
Second suggestion
When you use a numerical scale, be consistent with the meanings or ratings you assign to numbers. In the above example, "10" is excellent and "1" is poor. If you have another question on the questionnaire with a similar scale, ensure that "10" is excellent and "1" is poor. When you reverse the order (e.g., "1" is excellent and "10" is poor), you would create confusion among respondents.
In everyday conversation, we would call something excellent or outstanding a ten. To make it easy for respondents to understand the answering categories, choose descriptions that they are familiar with. It would reduce the chance of misinterpretation.
Third suggestion
Experiment with four answering categories for questions on agreement and intention. You would be surprised by how well four simple categories work.
To what extent do you agree with the statement that "breakfast is the most important meal of the day?" Would you say you ....
Strongly agree
Moderately agree
Moderately disagree
Strongly disagree
Here is an example of a question on purchasing intention:
As a result of attending the product presentation, how likely are you to buy Product A? Would you say you are ....
Very likely
Somewhat likely
Somewhat unlikely
Very unlikely
Subgroup analysis
It will help subgroup analysis by using four answer categories. Each answer category is a subgroup. You can look at each individually.
For instance, you can identify the demographics of those who say "very likely" and of those who say "very unlikely." If there are distinct differences in their respective demographics, you have excellent information for developing marketing programs to cater to each subgroup.
In contrast, if you use a 10-point or 12-point scale, you might see data scattered all over. It is difficult to conduct any useful analysis, when you are dealing with 10 or more subgroups.
In summary
When you design a scale of measurement, you have to be mindful of several elements. For the answering categories to be effective in capturing data, they should be simple to understand and not too many to be confusing. You should also think ahead to determine how best the scale can generate data for subgroup and other analyses.
Copyright © 2000 Pragmatic Consulting
Top of page | Gap Analysis | Unsettling Market Research
Back to Articles
Mission | Service | About Us | Applications | What's New | Contact | Home