Likert scales are among the most frequently used instruments in questionnaire surveys. Because they are relatively simple to design and fairly straightforward to interpret, we tend to use them a lot in applied linguistics research. This post has some information that you need to know if you are using Likert-scales in your research project.
In this post, you will learn the following four things:
- How to correctly pronounce ‘Likert’;
- If it is better to use odd or even numbers of responses;
- What the best number of responses is;
- Why you should not use weighted averages when analysing Likert data.
Before we start, I will very briefly go over what Likert-scales are and what my assumptions are about you, the intended reader. This will help to ensure that we are sharing the same initial assumptions.
What is a Likert scale?
A Likert scale is a group of statements and predefined responses that measure the intensity of the respondents’ feelings towards the preceding statement. Each statement and the answers that go with it are called an ‘item’. The construct that an item measures is called a ‘variable’. Here’s an example of an item:
|Strongly Agree||Agree||Disagree||Strongly Disagree|
|I just love Mondays!|
A Likert scale typically has multiple items, all of which measure the same underlying construct (or ‘latent variable’). In the example below, the four items measure a latent variable of ‘garfieldness’.
|Strongly Agree||Agree||Disagree||Strongly Disgree|
|I just love Mondays|
|I am very fond of lasagna|
|I am afraid of spiders|
|I am quite uncomfortable at the vet’s office|
Who is this post is for?
When writing this post, the primary audience that I had in mind is students in an applied linguistics or language educationcourse, perhaps working towards the completion of their dissertation or similar projects. I will therefore assume that you are understand basic mathematics, such as calculating averages. I will also assume that you do not need to know much about the technical aspects of Likert measurement. Finally, I will assume that you are competent with using statistical software, so I will not cover any of that here.
That said, much of the information in this post is likely to be useful for people working in diverse fields where Likert measurement is used.
1. Lick, not Like
Likert scales were created by Rensis Likert, a sociologist at the University of Michigan. The proper pronunciation of his name is “Lick – uhrt”. The pronunciation “like – uhrt”, though common, is incorrect.
2. Getting even helps
A Likert item consists of a prompt and a set of responses, often ranging from Strongly agree to Strongly disagree. There are usually five responses for each item, but seven-item scales are also quite common. When using an odd number of responses, the midpoint is a ‘neutral’ option, such as “no opinion”, “neither agree nor disagree”, “not sure” or some phrase to that effect.
What’s wrong with an even number of options?
Providing respondents with an even number of options has some advantages, but there are also two somewhat important problems, at least for our purposes in language teaching and applied linguistics research.
Firstly, many respondents tend to avoid voicing extreme opinions or taking a stand on controversial topics. This means that respondents are likely to select a ‘safe’ choice at the centre of the scale if one is available, rather than reveal their ‘true’ opinion – a phenomenon called the central tendency bias. This is especially the case when respondents are conscious of power imbalances (e.g., students responding to a questionnaire designed by their professors or teachers engaging with university-based research).
A second potential problem with middle options is that they can be hard to interpret. While we might assume that it means something along the lines of ‘I have no strong views either way’, this may not be true of all respondents. For some respondents, for example, the ‘neutral option’ could mean that ‘I don’t care either way’; for others it may mean that ‘I have no knowledge of this’.
Is there a better way to do this?
We can avoid some of these problems by using items that have an even number of responses. In the following example, respondents are presented with four ‘true’ options, which encourage them to voice a positive or negative opinion. This response format is called a ‘forced choice’ or ‘ipsative’ item.
|Strongly agree||Agree||Disagree||Strongly disagree||Never tasted it|
|Fish fingers and custard taste great|
The table above shows an ipsative item. This contains four ‘proper’ responses under the statement, in order to force respondents to register some agreement or disagreement.
There is also an additional ‘opt-out’ option for those respondents who truly cannot respond, but the wording of the item and the layout discourage its unnecessary use.
Disclaimer 1: Whether you use a ‘neutral’ option or not will depend a lot on your research aims, and the power dynamics in your research context. You might want to read more about the pros and cons of adding a neutral option in this article by TalentMap.
3. Less is more
Some Likert items contain large numbers of possible response options (7, 9 or 10) to capture a variety of nuanced positions. While such scales seem quite sensitive and accurate, they are not always very helpful. For one thing, any benefit from large numbers of options is subject to the law of diminishing returns. From the 7-option format and upwards, the scales just become too cumbersome to use. At that point, any additional benefits are cancelled out by respondent fatigue, and reliability plummets. Secondly, a large number of options might compromise the analytical sensitivity of the scales, because respondents tend to interpret the scales in different ways: what I describe as “often” may mean the same, in absolute terms, as what you might call “sometimes”. This phenomenon is amplified when the number of potential responses is large.
When interpreting the data, Likert items with many potential responses can sometimes be helpfully condensed into fewer, more meaningful categories. If you have an item with seven or nine responses, but a small sample size, this could mean only a small number of respondents have selevted each option. This is problematic because small numbers of respondents often limit the effectiveness of certain statistical procedures. In such cases, it might make sense to group all the ‘positive’ and ‘negative’ answers together. Doing so would involve the loss of some analytical detail, but this is an imperfect universe…
4. The mean is meaningless
The most common mistake in interpreting Likert scale data is reporting the mean values for responses. I have ranted about this practice elsewhere, but here’s the gist: To facilitate coding or save space on a questionnaire, we sometimes use numbers to represent response options in Likert items (e.g., Figure 1, top). These numerals are just descriptive codes, not ‘true’ numbers. From a mathematical perspective, a ‘Strongly Agree’ response indicates more agreement than ‘Agree’, but it does not show agreement that is five times stronger than ‘Strongly Disagree’. We could just as easily have used colours to anchor the responses, or any other symbol to show the same effect (e.g., Figure 1, bottom). In other words, we can use the data from Likert items (ordinal data, to be technical) if we want to rank responses, but that’s about the limit of what we can do with them .
Is it such a bad thing to calculate means?
To make this even clearer: We would be very unlikely to say that ‘the average response is agree and three quarters‘. Using numbers to express the same idea makes no more sense. Similarly, when we describe the fruit on a grocery stand, we can say that strawberries are smaller than apples, which are smaller than watermelons, and we can count how many fruit of each type are on sale, but we would never say that ‘the fruit on display are, on average, apples’. Reporting that the average of two agreements and strong disagreement is a ‘plain’ disagreement is just as bizarre.
Once more: when it comes to analysing the data that Likert items produce, reporting the mean makes very little mathematical sense (I am being charitable: others have called it an ‘indefensible‘ practice, and one of the seven ‘deadly sins‘ of statistics).
So what should one do instead?
The following set of posts contains some advice on how to analyse and interpret Likert scale data. The gist is that the safest metric of central tendency to use is the median. The mode is also a safe, but less useful metric.
For similar reasons, when we want to estimate the spread of responses in a Likert scale, it is best to use Range and InterQuartile Range (IQR). The Standard Deviation is a not a good choice, for reasons like the ones we have mentioned above.
It is also safer to avoid statistical procedures that rely on the mean (e.g. t-tests). Non-parametric tests, such as the Mann‐Whitney U-test, the Wilcoxon signed‐rank test and the Kruskal‐Wallis test are better alternatives.
For presenting data, it’s best to use bar charts, rather than histograms.
Here is some more advice about using Likert scales
Disclaimer 2: Under certain circumstances, a Likert scale (i.e., a collection of Likert items) can produce data that are suitable for calculating means, or running statistical tests that rely on the mean. These can be called ‘ordinal approximations of continuous data’. Experienced statisticians can probably get away with this, and they might be able to argue convincingly why their approach was appropriate. But if you’re doing a student project, the conservative approach suggested here is safer.
Additional reading about Likert scales
The advice and opinions in the previous sections were written to help you use Likert scales more effectively in your research projects. It has not been my intention to create an authoritative or comprehensive research methods guide, and I strongly encourage you to follow up on some of the things that you’ve just read. Some more resources that you may find helpful include the following:
- Likert, R. (1932). A technique for the measurement of attitudes. Archives of Psychology, 22(140).
- Cohen, L., Manion, L., & Morrison, K. (2000). Research methods in education (5th edn, pp. 253-255). Routledge.
- Gilbert, G. N. (2008). Researching social life (3rd edn, pp. 212ff.). SAGE.
Limitations of Likert scales
- Jamieson, S. (2004). Likert scales: how to (ab) use them. Medical Education, 38(12), 1217-1218.
- Matell, M. S., & Jacoby, J. (1971). Is there an optimal number of alternatives for Likert scale items? Educational and Psychological Measurement, 31(3), 657-674.
- Jacoby, J., & Matell, M. S. (1971). Three-point Likert scales are good enough. Journal of marketing research, 8(4), 495-500.
Some different views about Likert scales
The articles listed below describe perspectives on Likert scaling that are not in line with the recommendations I have made above.
- Norman, G. (2010). Likert scales, levels of measurement and the “laws” of statistics. Advances in Health Sciences Education, 15(5), 625-632. [This is a ‘rogue’ article, where the argument is made that, despite what purists claim, parametric procedures are robust enough to yield usable findings even when fed with ordinal (i.e., Likert-type) data.]
- Sullivan, G. M., & Artino, A. R. (2013). Analyzing and interpreting data from Likert-type scales. Journal of Graduate Medical Education, 5(4), 541–542. [This article extends the argument put forward by Norman (above). The authors concede that parametric tests tend to yield ‘correct’ results even if their assumptions are violated, but point out that “means are often of limited value unless the data follow a classic normal distribution and a frequency distribution of responses will likely be more helpful”. ]
Before you go: If you’ve landed on this page while preparing for a student project, I wish you good luck with your work. I hope that this information was helpful, but if there’s anything that was not clear, feel free to drop a line in the comments below or send me a message using the contact form. Also, please feel free to forward this information to anyone who might find it useful.
Achilleas Kostoulas is an applied linguist and language teacher educator. He teaches at the Department of Primary Education at the University of Thessaly, Greece. Previous academic affiliations include the University of Graz, Austria, and the University of Manchester, UK (which is also where he was awarded his PhD). He has extensive experience teaching research methods in the context of language teacher education.
About this post: This post was originally written in September 2013, based on lecture notes from a research methodology seminar that I was teaching at the time. It was last updated in April 2023. The featured image is by Michael Kwan and is shared under a Creative Commons licence. [CC BY-NC-ND]