Most online surveys contain rating scales in one form or another. These scales are where you ask people to choose from a range of options on a scale between two extreme values. They’re useful because they allow you to ask quantitative questions about abstract concepts, such as customer satisfaction, preferences or feelings, or experience of a product or service. Done well, they can help reduce the risk of your surveys containing bias and affecting your results. But one rating scale is not like another—they come in all shapes and sizes.
Curious to explore the other types of question formats we offer? Check out our question overview page to learn more!
There are many different types of rating scales, which suit different types of questions or situations. Firstly, there are ordinal (word-based) scales and interval scales (number-based or numerical rating scale). Then there’s a third type—the graphic rating scale—which can use either numbers or words. We’ll define each type below, talk about their advantages and disadvantages and look at some practical examples.
An ordinal scale is one where the answer options are expressed in words. For instance, you could be asked to rate your satisfaction level on a scale from “very satisfied” at one end, to “very unsatisfied” at the other. They normally contain a maximum of seven answer options as more than seven worded options can be confusing and overwhelming. Because these scales use words, the distance between the different options is not quantifiable. For instance, you can’t quantify the difference between “very satisfied” and “somewhat satisfied”. The well-known Likert scale[HL1] is a type of ordinal scale. Likert scales are often called satisfaction scales and normally include an odd number of potential answers, with one neutral option in the middle.
One of the biggest advantages of ordinal scales is that, since the different options are defined, it’s clear how each answer will be interpreted. Another benefit for survey organisers is that you can use the answer options when you report your findings. For instance, based on your survey results, you might report that “80% of Tesco customers were satisfied or very satisfied with their overall online shopping experience”.
A drawback to the ordinal, or word scale, is that people who aren’t fluent in the language may find the answer options difficult to understand. Ordinal scales also force survey respondents to choose from categories selected by the survey organiser. And this is a potential source of bias. The best way to overcome this issue is to thoroughly test your survey before you send it out, and to make sure you include an opt-out category like “N/A,” “Not Sure” or “Don’t Know.” Additionally, the fact that ordinal scales normally have a maximum of seven different answer options means they’re less precise than interval scales.
Interval scales include a series of numbered options, such as a scale from 1 to 10. These scales tend to be larger than ordinal scales, with more options to choose from. As the different options are numbered, they’re often known as numerical or numbered scales. With this type of scale, the intervals between each option are quantifiable. Text normally accompanies some of the numbers, at least each end of the scale. This way, the people answering will know what each end of the scale represents—for instance, is 1 “amazing” or is it “terrible”?
Interval scales’ greatest strength is their simplicity. They’re easy to understand and widely used around the world. Another advantage is that numerical rating scales are easy to analyse. You can crunch the numbers almost any way you want. And you can do all this straight away, since you don’t need to code in what each response represents (as with ordinal scales). Lastly, they give you more precise, granular data.
What does a 5 mean on a scale of 1 to 10? It’s a little subjective. Just as some people see a glass as half-empty, and others as half-full, people perceive numerical answers differently. This means that people with similar opinions may select different categories, since their perception of each category differs. This is a potential source of response error that makes it difficult to qualify what the data truly represents. Adding in descriptions of what each end of your numbered scale represents, either in the question or on your scale, can go some way towards solving this problem.
Graphic scales are often used to evaluate staff performance. They’re typically depicted in a table or matrix, with a list of qualities or traits down the left-hand column, and a series of rating options along the top row. The rating options can either be numbers, such as 1 to 5, or series of worded categories.
A key advantage of graphic scales is that they’re easy to understand. They’re also quick to answer. However, they do have their drawbacks. For a start, one evaluator may be harsher or more generous than another. And the “halo” effect can also mean that a strength in one area can result in overly generous rating in other areas.
How much do you agree or disagree with this statement? “I found it easy to transfer to First Direct.”
How satisfied were you with the service you received from HMRC today?
How would you rate your experience at the leisure centre today?
The Net Promoter Score (NPS) is a perfect example of an interval scale question.
For instance, John Lewis might ask its customers:
How likely would you be to recommend John Lewis to a friend or colleague?
Or Boots might ask online shoppers:
On a scale of 1 to 10, where 1 is dreadful and 10 is excellent, how would you rate your experience on our website?
Here’s an example of a graphic scale used in a staff appraisal.
Rate the performance of Alex Alphabet during the review period (September 2019 to August 2020)