Glossary

Likert Scale

A Likert scale is a symmetric, ordered response scale commonly used in surveys to measure attitudes or opinions. A typical five-point Likert scale ranges from "Strongly Agree" to "Strongly Disagree" with a neutral midpoint.

Definition

Why It Matters

Likert scales are the workhorse of survey research in the social sciences, market research, and organisational assessment. They offer a simple, standardised way to capture ordinal judgements, making it easy for respondents to answer and for researchers to aggregate responses. However, the ordinal nature of Likert data means that treating responses as interval-level measurements — for example, computing means and running t-tests — is technically controversial, though widely practised. Researchers should consider whether non-parametric methods or ordered logistic regression are more appropriate, especially with small sample sizes or skewed response distributions.

Example

A university administers a teaching evaluation questionnaire with 10 five-point Likert items (1 = Strongly Disagree to 5 = Strongly Agree). A course on statistical methods receives an average item score of 4.3 out of 5. While the mean is informative for benchmarking, ordinal logistic regression is used to model the probability of a top-box response as a function of student characteristics, respecting the ordinal nature of the data.

Related Terms

Software Notes

SPSS: Analyze > Descriptive Statistics > Frequencies for item-level summaries; Analyze > Regression > Ordinal for ordered logistic regression
R: table() and prop.table() for summaries; polr() from the MASS package or clm() from the ordinal package for ordered logistic regression
Stata: tabulate item for frequency tables; ologit y x1 x2 for ordered logistic regression