Read this topic

Researchers needed a fast way to turn a person's subjective agreement or disagreement into a number they could analyse, and the Likert-type scale answers that need. For example, a mood questionnaire might ask a patient to rate 'I feel hopeful about the future' from 'strongly disagree' to 'strongly agree' on a five-point scale. The scale connects to a bigger statistical problem: ordered categories are not the same as measured quantities, and treating them as if they were can mislead an analysis.

A Likert-type item presents a statement with an ordered set of response options, usually five or seven points, running from strong disagreement through a neutral midpoint to strong agreement. Each option is ranked relative to the others, so the data are ordinal: option 4 reflects more agreement than option 2, but the gap between points is not guaranteed to be equal. A summed or averaged score across several Likert items, a Likert scale as opposed to a single Likert item, behaves more like continuous data in practice. That is why many researchers apply parametric statistics to a total score while still treating a single item as ordinal.

The stem describes a questionnaire item with an ordered set of response labels running from 'strongly disagree' to 'strongly agree' (or similar), often five or seven options, and asks which statistical property or correct analysis applies.

Distinguish a single Likert item (ordinal, one statement, one rating) from a Likert scale (a multi-item total score, often analysed as approximately interval-level). Distinguish it from a visual analogue scale, which asks the patient to mark a point on a continuous line and produces genuinely continuous, interval-like data rather than discrete ranked categories.

Trap, assuming a five-point Likert item is interval data and computing a mean as if it were. Resolution: a single Likert item is ordinal, so report the median or mode, and use non-parametric tests unless working with a summed multi-item scale score. Trap: assuming the neutral midpoint always signals indecision. Resolution: some respondents pick the midpoint to avoid a socially undesirable extreme answer, a form of central tendency biasA systematic error that skews a study's results away from the truth in a particular direction.Read more →, not genuine neutrality.

Whenever a question describes ordered response categories, first ask: is this one item or a summed multi-item total? A single item is ordinal (median, non-parametric tests); a multi-item total is usually treated as approximately interval (mean, parametric tests) if the scale has enough items and points.