Likert Scales: 5-Point vs 7-Point, and When It Matters

Few methodology debates generate more heat and less consequence than 5-point vs 7-point Likert scales. The research literature's verdict, replicated many times: both are reliable, both are valid, and switching between them changes results far less than question wording does. But "it barely matters" is not "it never matters" - here is when it does.

Use 5 points when...

• •Your audience is the general consumer population - the extra granularity of 7 points mostly adds noise for casual respondents
• •The survey will be answered mainly on phones - five options fit a screen without scrolling; seven often do not, and options hidden behind a scroll get systematically under-picked
• •The survey is long - decision fatigue compounds with each extra option

Use 7 points when...

• •Respondents are experts or highly engaged (B2B panels, employees, physicians) who genuinely hold finer-grained opinions
• •You need to detect small movements between tracker waves - the extra points add sensitivity at the top of the scale where satisfaction studies live
• •You are averaging items into an index, where the extra granularity survives aggregation

The midpoint question

Include a neutral midpoint by default. Forcing a lean ("somewhat agree / somewhat disagree" with no middle) manufactures opinions that do not exist, and the manufactured leans do not replicate. Drop the midpoint only when the respondent definitionally must have a position - e.g. "which of these two would you choose?"

Labeling rules that matter more than length

1. Label every point, not just the ends - fully labeled scales are more reliable and harder to straight-line mindlessly
2. Keep the positive end on the same side for the entire survey
3. Keep the same scale length for the entire survey - mixing 5s and 7s poisons any cross-question comparison
4. Never number the points visibly unless the numbers mean something (like NPS's 0-10)

The defaults we recommend: 5-point, fully labeled, with midpoint, positive-right for LTR surveys and positive-left for Arabic RTL. And whichever you choose, the wording of the item will move your data more than the scale ever will.