How do you analyze Likert scale data?

First decide whether you have a single ordinal item or a validated scale that sums several items. Single items are summarised with the median, mode, and frequencies and tested with non-parametric methods, while a summated scale can often be treated as interval and analysed with means and parametric tests.

What is an example of a Likert scale analysis?

Comparing full-time and part-time students on a single agreement item is one example. You report the median and frequencies for each group, then run a Mann-Whitney U test to check whether the two distributions differ.

How to interpret 4 point Likert scale data?

A four-point scale forces a choice by removing the neutral midpoint, so responses are still ordinal categories. Summarise them with the median and frequencies, and compare groups with non-parametric tests such as Mann-Whitney U or Kruskal-Wallis.

What is the meaning of Likert scale analysis?

Likert scale analysis is the process of summarising and testing responses to agreement-style survey items. It covers describing the responses and choosing tests that respect whether the data is treated as ordinal or interval.

Analysing Likert-scale survey data

Likert-scale analysis is the set of methods you use to summarise and test responses to agreement-style survey items, the familiar "strongly disagree" to "strongly agree" choices. The crucial first decision is whether you treat each item as ordinal (ranked categories) or whether a validated summated scale of several items can be analysed as roughly interval. That single choice drives every test you run on your dissertation survey, so it is worth getting right before you touch the data.

Single item or summated scale: decide this first

A single Likert item is ordinal. The gaps between "agree" and "strongly agree" are not guaranteed to equal the gaps lower down, so the safest summaries are the median, the mode, and a table of frequencies, and the safest tests are non-parametric ones. A Likert scale, by contrast, sums or averages several items measuring one construct, and the resulting score is often treated as interval, which opens the door to means and parametric tests. Confirming the items hang together is a job for factor analysis and Cronbach's alpha before you compute a total.

Choosing the right test for your survey question

Once you know whether you have an ordinal item or an interval scale, the test follows from your research question. To compare two independent groups on an ordinal item, use the Mann-Whitney U test; for three or more groups, use the Kruskal-Wallis test. If you are working with a validated summated scale that behaves like interval data and meets the assumptions, an independent-samples t-test or ANOVA may be appropriate instead. The reasoning here mirrors how to choose a statistical test, and the ordinal-versus-interval split is exactly the territory of parametric versus non-parametric tests.

Start from the question your survey item answers, then follow it to the test family that respects whether the data is ordinal or interval.

A worked example you can copy

Suppose your survey asks one item, "I feel supported by my supervisor," on a five-point scale, and you want to compare full-time and part-time students. Because this is a single ordinal item, you report the median and the percentage choosing each option for both groups, then run a Mann-Whitney U test to see whether the distributions differ. If instead you had ten items measuring overall satisfaction, you would confirm reliability, sum them into one score, check the assumptions, and then compare group means with a t-test. Same survey, two different routes, decided entirely by single item versus scale.

Reporting Likert results in your dissertation

Present descriptive results first, a clear table of frequencies or medians, then your test result with its p-value and an effect size so the reader can judge the size of the difference, not just its significance. Diverging stacked bar charts read well for agreement items. The distinction between describing your sample and generalising from it, covered in descriptive versus inferential statistics, applies here just as it does to any other data. If you want the scale checks and the testing handled end to end, that is the work in dissertation data analysis help.