Written and maintained by CASRAI Editorial Board
Last updated
The Likert scale (developed by Rensis Likert in 1932) is the standard psychometric scale used in survey research, organizational psychology, and behavioral sciences to quantify latent attitudes, beliefs, and opinions. A major methodological debate centers on whether Likert data should be treated as ordinal (requiring non-parametric statistics such as Mann-Whitney U, Wilcoxon, Kruskal-Wallis, and medians) or interval (permitting parametric statistics such as t-tests, ANOVA, linear regression, means, and standard deviations). This guide outlines Likert response formatting, odd vs. even point formats, and appropriate statistical testing workflows.
Likert Item vs. Likert Scale: The Essential Distinction
- Individual Likert Item: A single survey question with response anchors (e.g., 1 = Strongly Disagree to 5 = Strongly Agree). A single item is strictly ordinal data; computing means and standard deviations on a single item violates measurement theory.
- Composite Likert Scale: The sum or average of multiple Likert items measuring the same underlying construct (verified via Cronbach’s alpha ≥ 0.70 and factor analysis). Composite scale scores exhibit continuous, interval-like distributions and can be legitimately analyzed using parametric statistics.
Response Point Formats: 5-Point vs 7-Point vs Forced-Choice (Even)
| Scale Format | Neutral Midpoint | Psychometric Precision & Reliability | Primary Research Use Case |
|---|---|---|---|
| 5-Point Likert | Included (Neither Agree nor Disagree) | Optimal cognitive clarity; minimizes respondent fatigue across long surveys. | General population surveys, mobile questionnaires, student evaluations. |
| 7-Point Likert | Included (Neutral) | Highest psychometric variance and sensitivity; closer approximation to continuous normal distribution. | Academic psychology research, validated psychometric scale development. |
| 4-Point / 6-Point (Even) | Omitted (Forced-Choice) | Eliminates neutral fence-sitting bias; forces directional cognitive commitment. | Employee engagement audits, controversial policy polling, marketing preference trials. |
