Likert scale – left or right?

January 14, 2014 at 9:15 pm 8 comments

I’ve written previously about the Likert  scale and using it in surveys. On one point, I discussed whether the response options should be displayed positive to negative or negative to positive – the image below is negative (‘strongly disagree’) to positive (‘strongly agree’).  

I’ve recently come across two articles (listed below) where they have researched this issue – and they have found that the place of the response option does matter.  In summary, they found that the items placed on the left hand side gets more people selecting them than those on the right. They also found that when using vertical lists, the first items are more selected than others further down the list.

So what is the solution? One option I see is that many survey software offer the possibility when creating questions for the response options to be “flipped” – that is, some people will see them negative to positive and others will see them positive to negative.  It also makes sense to vary the response order in long surveys, particularly when using same or similar scales – to avoid respondents suffering from “survey fatigue”.

The relevant articles:
Response order effects in Likert-type Scales (pdf) >>
The biasing effect of scale-checking styles on response to a Likert Scale (pdf)>>

Entry filed under: Evaluation tools (surveys, interviews..).

Evaluators who blog Workshop on evaluation and complexity – Berne, June 5-6 – registration open

8 Comments Add your own

  • 1. Ann Larson  |  January 15, 2014 at 1:32 am

    Thanks for the post. It is an issue that I ponder every time I conduct a survey. But I am not convinced that this is simply a matter of bias towards answering the left-hand responses. Buried in the results section of the first paper you hyperlinked is the comment that the scales that start with the positive categories are ‘easier’. That makes sense to me. When reflecting on a statement I think that a people (okay I am a glass half full girl) think ‘do I agree with this–yes or no’. If yes they tick the closest boxes if not they look further across the columns. It is not as natural to think ‘do I not agree with this’. In any event, the most important response for researchers may not be to randomly assign the order but to be consistent across surveys.

  • 2. Glenn  |  January 15, 2014 at 7:26 am

    Thanks Ann for that comment, I think also that if categories starting with positive are ‘easier’ – then it may make sense to start with the negative? I find that people are over positive in surveys and I’ve never seen an obvious bias towards negative responses. Also I agree it is important to be consistent across surveys but it’s important to find a balance between consistency and repetition, i.e. using only the same type of questions and response options can lead to “survey fatigue” where users click through all automatically.

  • 3. David Wookey  |  January 15, 2014 at 9:06 am

    Isn’t there evidence to suggest that people are more likely to respond in the middle 3 boxes and avoid extremities in responses (unless they are really trying to drive home a point). This is irrespective of the order that the responses may be in. Thus, the drive to use 7 or even 9 point scales to attempt to get more variation to the scale. Having managed employee surveys as well as design surveys for employee development, I’m interested to know if this is correct.

  • 4. Glenn  |  January 16, 2014 at 3:52 pm

    Hi David, that’s a good point – I haven’t seen any studies that show people avoid the extremities – but I’m ready to stand corrected if some are found… what I have read is that studies show that people are not able to place their point of view on a scale greater than seven. So seven or less is recommended. What is the perfect number? Studies are not conclusive on this, most commonly mentioned are five, four or three point scales. But certainly Mr Likert (yes, it’s named after him…) created this scale as he thought that a simple yes/no scale could not accommodate all possible shades of responses – and therefore number of responses depends upon what you are asking about – some issues may need up to 11 options – this article has some good information on this point:

  • 5. Pete  |  October 22, 2014 at 6:20 pm

    a lot of the discussion here seems to involve surveys, but i am interested in how this applies to self-assessments, which are more diagnostic than evaluative, and have no risk of bias towards positive or negative. I have noticed that users tend to want to click the right side of a scale to agree and the left to disagree, and i was wondering if there are any empirical studies that address this.

  • 6. Glenn  |  October 31, 2014 at 9:20 pm

    Hi Pete, I think you would have to dig further into the article referred to in the post – but I believe they will confirm that self-assessment forms are perceived by users the same as for surveys, i.e. if the scale is left to right, or right to left can influence the user.

    • 7. Fru  |  July 22, 2018 at 2:23 pm

      I think it is subjective and depends on the researchers and his objectives . If for example am trying to proof a point I can start with Strongly Agree to Strongly Disagree because Respondents will naturally select options closer to them

      Like when I was doing my research I wanted to show how Motivation impacted learning and so I started with SA to SD

  • […] Surveys: in the association industry, you likely have seen many number of surveys from membership feedback, conference evaluations, board and committee responses, and more. The Likert Scale is a useful survey format for getting responses, particularly if done properly. Do a quick google images search for Likert Scale. You will find many examples of the scale where the “positive” responses (strongly agree, excellent, etc) are on the left side of the scale, and many others where the “negative” responses (strongly disagree, poor, etc) are on the same left side. When designing a survey, which side should this be on? Is there a correct answer? It may not seem to matter, particularly when searching through all of the likert scales used in surveys and seeing a close to 50-50 split in this ordering. Nevertheless, research has been done to provide some feedback on this issue. Intelligentmeasurement has some results from research on this topic. You’ll find their findings interesting if you do a lot of surveys. Check it out here. […]


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

Trackback this post  |  Subscribe to the comments via RSS Feed

Enter your email address to subscribe to this blog and receive notifications of new posts by email.

Join 1,160 other subscribers



%d bloggers like this: