Date: Saturday, March 8, 2025
Hi! I am Thuy Dung Pham. I am a Postdoctoral Research Associate in Evaluation and Statistical Methods at University of Georgia Extension. In this blog, I would like to share my thoughts on how to analyze Likert-scale data.
As you may know, a Likert scale is a very common type of rating scale used in surveys. In Extension, we often use it to gauge someone’s perception on how well the information will help them engage in safer practices. For example, food science tells us that it is not safe to wash meat before cooking- this can spread bacteria up to 3-4 feet from the sink! A program to increase food safety will ask people how helpful this information will be in influencing a change in behavior. People may be asked to express their opinion by selecting one from five options: Extremely helpful, Very helpful, Somewhat helpful, Slightly helpful, and Not at all helpful (i.e., a 5-point Likert scale).
Another way Likert scales are used in Extension are to assess a concept. One recent example is using a validated scale to assess the level of stigma towards people with opioid use disorders, with responses ranging from 1 (Strongly disagree) to 5 (Strongly agree). Note that in the first example, we are to analyze one survey question while in the second example, we are to analyze a set of survey questions that has been validated to be presented together as a set and with the same response options across the set.
The question is- what is the best way to look at this data? In general, there are two approaches to analyzing Likert-scale data: treating them as ordinal or interval/continuous data. However, the best approach is often based on the situation.
Likert-scale data are generally considered ordinal, meaning the ordered categories may not be equally spaced. For instance, the difference between “Extremely helpful” and “Very helpful” is probably smaller than between “Very helpful” and “Somewhat helpful”. When we are to analyze only one survey item as in the first example, I think treating it as an ordinal variable is the way to go. When analyzing multiple items as a set, and if it is appropriate to combine them into a single scale score (and/or subscale scores if the scale is made up of subscales), as in the second example, one may consider treating the scale/subscale scores as interval/continuous variables.
So what does this mean in practice? If you are treating the data as interval, you can use parametric tests. A parametric test that is often used in Extension is t-test, which compares differences in the means between two (independent or dependent) groups. Another common parametric test is ANOVA, which compares differences in the means between three or more groups. In this case, reporting the mean and standard deviation is the appropriate way to summarize the data.
If you are treating the data as ordinal, the median, which represents the middle value in a data set, is the appropriate way to summarize the data. When it comes to statistical tests, non-parametric tests are more appropriate for ordinal data in most situations. Some non-parametric tests that compare differences (in the medians) between groups include:
The American Evaluation Association is hosting Extension Education Evaluation TIG Week with our colleagues in the Extension Education Evaluation Topical Interest Group. The contributions all this week to AEA365 come from our EEE TIG members. Do you have questions, concerns, kudos, or content to extend this AEA365 contribution? Please add them in the comments section for this post on the AEA365 webpage so that we may enrich our community of practice. Would you like to submit an AEA365 Tip? Please send a note of interest to AEA365@eval.org. AEA365 is sponsored by the American Evaluation Association and provides a Tip-a-Day by and for evaluators. The views and opinions expressed on the AEA365 blog are solely those of the original authors and other contributors. These views and opinions do not necessarily represent those of the American Evaluation Association, and/or any/all contributors to this site.