Likert scales are an example of a summated scale, something which is very rare in market research and quite common in social and academic research.
The classic use of a Likert scale is to measure where people lie on some underlying scale. For example, we might believe that people rate products on three underlying dimensions, for example Evaluation, Potency, and Activation.
In order to measure where somebody places a product on these three dimensions we might develop 10 Likert items for each dimension. For example, on question/item might be ’To what extent do you agree that this product works within 2 hours?’ as an item for the Potency dimension. The Likert item tends to use language such as the traditional Agree Strongly to Disagree Strongly. In best practice, half the Likert items will be worded positively, half negatively, and the scores reversed when assessing the negatively worded questions.
All of the Likert items being used for a single dimension should be highly correlated in the sense that they measure the dimension and do not measure something else. Some of the statements will be strongly worded, such that relatively few respondents will agree, whilst others are worded so that many people will agree – to create a more sensitive scale.
In this example, after the respondent has entered their replies for all 30 Likert items (and the negatives reversed) the three Likert scales are calculated by adding together the numbers from the Likert items. So, in this example, the Likert scale for Evaluation would range from 10 (where the respondent disagreed strongly with each Likert item) to 50 (where the respondent strongly agreed with each Likert item).
Summated Likert scales are typically validated against theory and alternative measurements to show that they measure what they purport to measure.
Market Research 5 point rating scale
Market research often uses scales that look like Likert items, but they are not Likert scales (although this does not stop some market research textbooks calling them Likert scales). The key differences are:
Likert scales are based on questions that are deliberately highly correlated, market research questions often seek to minimise correlations between attributes.
Likert scales combine several questions (items) together to make a single score. Market research typically reports each question as an item.
In market research, rating scales that are superficially based on Likert items are often criticised because there is no basis for their treatment of the 1 to 5 values as an interval scale (if we want to calculate means they need to be at least interval scales). They are also criticised because when they are misused to generate means they tend to produce values with little variation.