3. Explore Various Rating Scales
3.2 Numerical rating scales
Numerical (or numeric) rating scales are designed with numbers (integers) along with a few verbal labels or visual cues. They can be designed to measure various aspects—such as levels of change, satisfaction, or quality—using different ranges of numerical values (e.g., an 11-point scale from 0 to 10, a 7-point scale from -3 to +3, or a 5-point scale from 1 to 5).
These scales are typically ordinal or interval in nature and are considered discrete rating scales, as they produce specific integer values as data. For example, with a 5-point numerical rating scale, respondents can select only one of the five numbers, generating values like1, 2, 3, 4, or 5. Numerical rating scales are not intended to record data in fractional numbers such as 1.24, 3.26, or 4.38.
Numerical rating scales are usually partially labeled, with only the two endpoints labeled, as shown below.
Poor 1 2 3 4 5 Excellent
Unlikely 1 2 3 4 5 6 7 Likely
Very dissatisfied -3 -2 -1 0 +1 +2 +3 Very satisfied
A type of numerical rating scale, known as the semantic differential, consists of a set of partially-labeled N-point scales with two opposing end labels. Typically, 5-point or 7-point scales are used. The bipolar adjectives at each end describe contrasting traits of the object being measured. For example, a semantic differential scale designed to measure a trainers’ performance might include the following set of opposing descriptors:
Please rate the trainer’s performance:
| [negative] | 1 | 2 | 3 | 4 | 5 | 6 | 7 | [positive] |
|---|---|---|---|---|---|---|---|---|
| Unready | ○ | ○ | ○ | ○ | ○ | ○ | ○ | Prepared |
| Distant | ○ | ○ | ○ | ○ | ○ | ○ | ○ | Friendly |
| Inexperienced | ○ | ○ | ○ | ○ | ○ | ○ | ○ | Knowledgeable |
| Amateur | ○ | ○ | ○ | ○ | ○ | ○ | ○ | Professional |
| Rude | ○ | ○ | ○ | ○ | ○ | ○ | ○ | Respectful |
| Terrible | ○ | ○ | ○ | ○ | ○ | ○ | ○ | Outstanding |
Numerical rating scales may include more than two ending labels. As shown below, some provide one or more intermediate labels between the endpoints, to offer more context for each response option. Numerical rating scales can be unipolar or bipolar. The first two examples below illustrate unipolar scales, while the third example shows a bipolar scale.
| 0 No pain |
1 | 2 | 3 | 4 | 5 Mod- erate pain |
6 | 7 | 8 | 9 | 10 Worst pain ever |
| 0 No change |
1 | 2 A bit |
3 | 4 | 5 Some |
6 | 7 | 8 Much |
9 | 10 Most possible change |
| 1 Strongly disagree |
2 | 3 | 4 Neither |
5 | 6 | 7 Strongly agree |
Numerical rating scales are relatively easy to develop and use; however, they should be applied with caution. Descriptive labels added to numerical rating scales assist respondents in selecting a number on the scale that best represents their experience or opinion. However, partially-labeled numerical rating scales leave room for individual interpretation, which can vary widely among respondents.
For example, let’s say, Respondent A and Respondent B both have the same level of mild pain from the same symptom, but Respondent A may choose 1 and Respondent B may choose 2 or 3 on the 11-point pain-measuring scale shown above.
This variation may lead you to consider fully labeling each point on a numerical rating scale, as shown below:
| Amount of difficulty |
1 None |
2 A bit |
3 Some |
4 Quite a bit |
5 A lot |
|---|
However, fully-labeled rating scales are better categorized as verbal descriptor scales rather than numerical rating scales, because now respondents are making their selections based primarily on the descriptive labels rather than the numbers. In these cases, the numbers (e.g., 1 through 5) serve mainly as coding values for data analysis and can even be removed from the scale entirely, as shown below. More information about verbal descriptor scales is presented in the following section.
| Amount of difficulty |
○ None |
○ A bit |
○ Some |
○ Quite a bit |
○ A lot |
|---|
