Forms · Glossary

Should a survey scale have a neutral option?

A neutral option is the midpoint of a scale, usually labelled neither agree nor disagree or neutral. Including it lets people with a genuinely balanced view say so. Leaving it out, which produces a forced choice scale with an even number of points, pushes everybody to one side or the other.

The argument for removing the midpoint is that it collects people who could not be bothered. The argument for keeping it is that it also collects people who genuinely have no view, and removing it does not give you their real opinion, only a guess.

Nuwan Madhusanka · Co-founder

6 min read · Published 12 September 2026

Odd against even numbers of points
	5 point, with a midpoint	4 point, forced choice
Respondent with no view	Chooses the midpoint	Guesses, or abandons the question
Respondent with a weak view	May park in the middle	Records the direction of the weak view
Interpreting the middle	Ambiguous: no view, mixed view, or no effort	No ambiguity, because there is none
Missing data	Lower, people have somewhere to go	Higher, especially on sensitive items
Reported as	Distribution, or top two box	Share positive against share negative
Best for	Most audiences, most subjects	Forced prioritisation, expert raters

What the midpoint actually collects

This is the heart of the argument, because the middle position is not one thing. It holds people who genuinely feel balanced, people who have a mixed view with strong feelings on both sides, people who lack the knowledge to judge, and people who are moving through the survey quickly. Those four groups need different responses from the organisation and the scale cannot tell them apart. That ambiguity is the strongest argument against the midpoint, and it is also an argument for a different fix: separating the reasons. Offering a not applicable or a do not know option alongside a five point scale removes the group with no knowledge from the middle, which makes the remaining midpoint far easier to interpret.

What happens when you remove it

Published comparisons of four and five point versions of the same items consistently find that removing the midpoint does not simply split those respondents evenly across the two sides. Some choose a side, some choose the nearest position to the middle, and some skip the question or abandon the survey, which raises missing data. The direction of the split also tends to favour the positive side on questions about a named organisation, because agreeing remains the smoother answer. That means a forced choice scale does not reveal a hidden opinion so much as manufacture a weak one, and the manufactured answers are indistinguishable from real ones in the data.

When forced choice is the right call

There are cases for it. When the respondents are experts rating something they know well, a midpoint is more likely to be evasion than genuine balance. When the purpose is to force a prioritisation, such as choosing which of two options to build, sitting in the middle defeats the question. And in some performance or clinical rating contexts, a forced choice is used deliberately to prevent central tendency, where raters cluster everything near the middle to avoid committing. Outside those situations, the default for a general audience should be an odd number of points with the midpoint labelled clearly and a separate escape option for people who cannot judge.

Labelling the middle so it means something

An unlabelled midpoint invites every interpretation there is. Label it, and choose the label carefully, because neutral, neither agree nor disagree, and no opinion mean different things. Neither agree nor disagree describes a balanced view; no opinion describes an absence of one, and mixing them into a single option merges two groups again. If your items are item specific rather than agreement based, which is generally better, the midpoint gets easier to label: a scale running from not at all useful to extremely useful has a natural middle of moderately useful, which carries no ambiguity at all. That is one more reason to prefer item specific scales over agreement ones.

Building the scale in a form

A five point scale is a radio field with five options, each labelled in words, laid out in one, two or three columns. Where several statements share the scale, a matrix is the compact build: rows by columns, with radio style allowing one answer per row. If you add a separate not applicable option, it belongs as an extra column in the matrix or an extra option on the radio field, placed at the end and visually separated in the wording. A matrix exports as one column with each statement and its answer joined as Statement: Answer, so plan to split that column before analysing any single item.

Deciding once and holding to it

The most costly version of this debate is the one that gets reopened every year. A team switches from five points to four because an executive dislikes the size of the middle group, then switches back when the forced choice version produces more missing data, and two years of trend become unreadable. Settle the question at the start of a measurement programme, write the reasoning down alongside the question wording, and treat any change as a deliberate break with a parallel wave to bridge it. It is also worth agreeing in advance how the midpoint will be reported, because a large middle group looks like a problem to anybody seeing it for the first time. Explaining beforehand that a balanced view is a legitimate answer saves the conversation that otherwise leads to a scale change nobody needed.

Questions people ask

Is a five point scale better than a seven point one?

For general audiences, five is easier to label in words and finishes faster. Seven gives more room to show movement and suits engaged respondents making fine distinctions. Both have a midpoint, so the question of whether to include one is separate from how many points to use.

Does the midpoint attract lazy answers?

Some, and the honest response is to reduce the cost of thinking rather than to remove the option. Shorter surveys, fewer statements per grid, and item specific scales all cut the number of people who park in the middle. Removing the midpoint moves those people rather than engaging them.

How should I report a midpoint in results?

Show it. Collapsing a five point scale into positive and negative and dropping the middle hides a group that may be the largest one. Reporting the top two points, the middle and the bottom two keeps the shape visible, and a shift out of the middle in either direction is often the most interesting movement in a tracking survey.

Should a midpoint be a number or a word?

A word, in the same register as the other labels. A scale where the ends are labelled and the middle is a bare three leaves people to infer what three means, and they infer differently. Labelling every point costs nothing and makes answers comparable between respondents and across waves.

What if most answers land in the middle?

That is usually a signal about the question rather than the scale. It happens when the item is abstract, when the respondent has no basis to judge, or when the subject genuinely does not matter to them. Check whether a not applicable option would drain the middle, and consider whether the item is worth asking at all.

Can I change from an even to an odd scale mid series?

You can, and it breaks the comparison with everything measured before. If the change is necessary, run both versions on split halves of the sample for one wave so the shift can be measured, then report the break clearly. A quiet switch is the version that produces a misleading trend line nobody can explain later.

Make one with forms

The button opens the generator with this use case already described. Change the wording to match your own.

Create a form with OneCraft

Sources

Written and checked by the OneCraft team. Last checked 12 September 2026.