Rating scale

"Rate from 1 to 5", "Give 1 to 5 stars", "How likely on a 0-10 scale" - these are all rating scales. The respondent picks a value on an ordered scale: numbers, stars, emoji, or labels.

The difference from the Likert scale is that Likert measures agreement with a statement, while a rating scale is broader: you can rate anything - quality, likelihood, importance. In SurveyNinja it is built with the "Scale" element, where you configure labels and the display type. Below is when to use which scale and what to watch for when interpreting the results.

A rating gives you a number or category - handy for computing the mean, median, and shares at the extremes. But the points are ordered, not necessarily equally spaced: the "gap between 4 and 5" may be perceived differently from the "gap between 1 and 2". For practical purposes the mean is often used, but a strict interpretation like "0.5 points higher" is not always correct.

Numbers, stars, emoji - types of rating scales

Rating scale - a measurement method in which the respondent picks a value on an ordered scale. Variants: numeric (1-5, 1-10), stars, emoji (smileys), or verbal labels ("poor" - "excellent"). Used to rate quality, satisfaction, likelihood, or importance. The data is ordinal (or pseudo-interval with 5+ points). The mean and median are the typical metrics; for comparing segments and survey waves, a rating is convenient and easy to grasp.

In short: "pick a value from X to Y" - and the respondent clicks the point they want.

Rating vs the Likert scale

Likert is a reaction to a statement along degrees of agreement ("agree - disagree"). A rating scale is the umbrella term: rate anything along ordered points. "Rate quality from 1 to 5" is a rating. "Do you agree that the quality is high?" with agreement options is Likert. In practice a "1-5 rating" is often called both; strictly speaking, rating is the broader term. More on this in the article about the Likert scale.

5, 7 or 10 points - which to choose

5 is the most common option: familiar and not overwhelming. 7 captures more nuance, but the points can start to "blur together". 10 is familiar from NPS and "out of ten" ratings; on mobile, 10 buttons take up a lot of space. 3-4 is coarse but fast. The choice depends on the task: for comparison against benchmarks it is sometimes important to keep the same scale (for example, always 1-5).

An odd number gives a neutral midpoint; an even number forces a leaning. "Hard to say" can be added separately, but then it sits outside the scale - as a separate option.

Stars, emoji, numbers - impact on answers

Numbers are neutral and suit any topic. Stars are associated with ratings and reviews; they are familiar from feedback. Emoji (smileys) feel more emotional and are sometimes used in CSAT and surveys for children. The visual format affects perception: stars can bias answers toward the positive end. For serious research, numbers or verbal labels are more common.

In SurveyNinja: the "Scale" element, labels, display type

A rating scale in the SurveyNinja builder is the "Scale" element. You set the number of points (for example 5), the labels (1-5 or "poor" - "excellent"), and the display type - numbers, stars, buttons. You can make it required. Details are in the element settings.

Answers in the reports show the distribution across points and the mean. The Excel export gives a column with the chosen value. It is easy to build cross-tabulations by segment - average rating of men vs women, by region, and so on.

Mean, median, shares - what to look at

The mean is simple and clear. But with a skewed distribution (many "5"s and few "1"s) the median is more robust. Shares at the extremes - the "share who rated 4-5" - are often more telling for a client than "a mean of 4.2". Look at both: the mean for comparison, the distribution for understanding.

With a small sample (N<30) the mean on a 5-point scale can swing a lot from a couple of extreme answers. The median and shares are more stable.

Areas of application

Rating the quality of a service, product, or event. The importance of attributes in product research. The likelihood of an action ("How likely are you to recommend?"). Satisfaction with various aspects. Anything that needs a numeric or ordinal value instead of free text - a rating scale fits. Its connection to the scale in the broad sense - a rating scale is a special case of it.

Common mistakes

Too many points. 10-11 points make it hard to tell neighbors apart. 5-7 is usually enough.

Unclear poles. "1 - minimum, 10 - maximum" without context - the minimum and maximum of what? Better to use labels: "1 - not at all important", "10 - extremely important".

Mixing directions. Having "5 = excellent" and "5 = poor" in the same block will confuse respondents - and you, during analysis. One pole, one meaning.

Treating it as an interval scale. "The mean rose by 0.3" is fine for a trend. "A difference of 0.3 is statistically significant" needs to be tested. The points are ordinal; a strict "so many points" interpretation comes with caveats.

NPS, CSAT, overall rating - where rating fits

NPS (0-10) is a rating scale with a special calculation (shares of promoters/passives/detractors). CSAT is often 1-5 or smileys. An overall rating of a service, product, or event is a typical 1-5 or 1-10 rating. In quantitative surveys, rating is a basic building block: fast, clear, and easy to analyze.

Case: rating a webinar 1-5

After a webinar: "Rate from 1 to 5: relevance of the topic, quality of the presentation, answers to questions". Three rating questions (or a single matrix). 150 responses: means of 4.3, 4.1, 3.9. "Answers to questions" is lower - a zone for improvement. The "4-5" share is 78%, 72%, 65%. For the report to the sponsor - both the means and the shares. The rating gave compact metrics without any coding.

Slider vs buttons - which is more convenient

A rating can be shown with buttons (1, 2, 3, 4, 5) or a slider. Buttons are an explicit choice - the respondent sees all the points. A slider is compact, but on mobile its precision is lower (easy to miss), and not everyone understands which way to drag. For 5-10 points, buttons are more reliable. A slider is appropriate for continuous ratings or when there are many points - but then it is already a different question type.

Document the scale in your methodology

In the report and methodology, state: which scale (1-5, 1-10), the labels for the poles, and the question wording. This is needed for reproducibility and comparison with other measurements. "A 5-point scale from 1 (poor) to 5 (excellent) was used" is enough for most cases.

A rating scale - rating on ordered points (numbers, stars, labels). Broader than Likert - you can rate quality, importance, likelihood. In SurveyNinja it is the "Scale" element with configurable labels and display type.