Survey Question Writing Guide: Question Scale Length

writing scale questions

When you get right down to it, there are two purposes of a survey scale question: learn what people think and distribute answers along a continuum.

The second part allows us to compare answers individually (response A versus response B) or as a group (25 year-old buyers versus 20 year-olds).

Using scale questions that offer a complete spectrum of options lets respondents give their true opinions while reducing the length of the survey, so it’s important to get these survey questions right.

To make sure you’re asking the right questions in the right way, we’re going to investigate the pros and cons of various scale lengths and sizes.

Ideal Length for Scale Questions: 5 or 7 Options

For the best data, we recommend using an odd-numbered scale. This allows respondents to select a neutral option and gives a reasonable number of choices on both the positive and negative sides of the scale.

If you expect your results to be fairly equitable, meaning fairly equal numbers of positive and negative responses, then a 5-point scale will be your best bet.

5 point scale survey question

If, on the other hand, you suspect that you’ll be getting responses that skew to one end of the spectrum or the other, you should opt for the 7-point scale.

7 point survey scale question

Want to learn more about these kinds of survey questions?
Download our free ebook: A Guide to Quantitative Questions.

Download the Ebook

The Trouble With Even-Numbered Scale Questions

It can be tempting to create scale questions with an even number of responses like the one below, because these force survey takers to choose either a positive or negative answer.

4 point survey question scale

But while these kinds of questions can actually impoverish your survey data by forcing people who are truly neutral or undecided to arbitrarily choose a side.

This can be frustrating to a respondent who doesn’t feel that their input is being accurately represented, and it can be detrimental to the survey’s accuracy as well.

If you’re skeptical about the impact of using only odd-numbered scale questions, there’s a simple way to test them. Ask the same question in two different parts of your survey, once using an odd-numbered sale and once using an even-numbered scale. Compare the answers, and you’ll almost certainly find that the results are comparable if not statistically identical.

Scale Question Length: How Long is Too Long?

There’s a reason that telephone numbers are only seven digits long. The human mind can easily process approximately seven data bits before it gets overwhelmed and has trouble remembering.

In survey questions, this means that we need to limit the number of scale options to seven or fewer, otherwise respondents will struggle to process the scale itself.

More than seven levels of agreement cannot be considered at once, which makes it difficult to respond to scales with more than seven levels. In short, they are very fatiguing. They deprive the mind of the chance to embrace the scale as one and accurately make a selection from an array of balanced alternatives.

What the mind does with a ten-point scale, for example, is first split it into a positive and negative half. This goes back to the issue with even-numbered scales: the even split makes it frustrating for respondents who want a neutral option.

Scale Question Summary

In sum, using scales that provide a full array of alternatives within an appropriate length permits respondents to complete your survey in a shorter amount of time and with greater accuracy.

  • 4-point scales: an impoverished an inaccurate data set
  • 5-point scales: accurate, and meets your needs 90% of the time
  • 7-point scales: accurate, and more useful for data that skews positive or negative
  • 10-point scales: very fatiguing, and should be avoided at all times
Join the Conversation
  • Tyler Totman

    What is your opinion of the “Net Promotor Score” method with 10 point scales?  For example:
    “How likely are you to refer our product/service to a friend?”
    1 – 10Then they use the following:    [ Sum (1-5) ] [ discard 6,7]  [Sum (8,9,10) ]  [Sum (8,9,10) ] minus [ Sum (1-5) ]  The result is your promoters less your attractors, for a solid measurement of how people think/promote your brand?

    I see how it could be modified for a 5 point scale (or 7), but not sure what that would do to effectiveness of the simple answer (especially if it is being used in a balance scorecard etc.) 

    • sgizmo

      Although Bill recommends the 7-point scale, Fred Reicheld, the author of The Ultimate Question, recommends the 11-point scale rather than a 10 point scale. The 11-point scale gives you the most variability for selecting the likelihood to recommend or not. The importance of the 0 is so a respondent that would not recommend a product or service at all has a valid answer to select. A 1 may indicate that they see some, although limited value in the product or service and may lead the respondent to not answer the question, thus tainting the overall data.

  • SS

    hi!  i want to ask government officials  how important “issue x” is for them.  Do you suggest using a 10 point scale (ie “how important is x issue” and they would pick a number b/n 1 and 10  or can i use a 7 point likert.  I guess more importantly, when can i use a likert scale and when i use a regular scale? 
    Thanks so much for your time.

  • Robin

    I don’t agree with you 100%. I think the value of practice is really important, but there is research out there that has demonstrated (given, in their limited settings) that a mid-point can bias results (see for example Garland (1991) The Mid-Point on a Rating Scale: Is it Desirable? – there are other more recent studies that I don’t have on hand). As a survey researcher myself, there have been contexts where it is absolutely necessary to force a choice, and others where, as you suggest, it makes no sense to do so. It is highly context dependent. I don’t see why you come so strong on one end of the spectrum. If you have empirical research (other than gut feeling or past experience in an uncontrolled environment) I think it is your responsibility as a researcher to cite that when giving technical advice to others.

    • sgizmo

      You are absolutely correct: there are reasons why a survey might benefit from including a neutral mid-point or not. It really depends on what the researcher is trying to accomplish. Thanks for pointing out that methodology can (and should) shift depending on a survey’s intent.