Help wanted -5-point Likert or 10-point numerical?

May 26, 2010 at 7:36 pm 7 comments

Here’s one for you our readers.

Benchpoint is currently designing a survey for a client. Most of the questions have 5-point Likert scales:

Very satisfied
Satisfied
Neither satisfied nor dissatisfied
Slightly dissatisfied
Very dissatisfied

However the client wishes to have one question with a 10 point numerical scale where 9 is extremely satisfied and 0 is extremely dissatisfied.

We say we should stick to the same scale throughout the survey, and that a 5-point descriptive scale is better that a 10-point numerical scale.

What do our readers think?

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

Web analytics and communications evaluation Researching campaign messages – wrinkles or cancer?

7 Comments Add your own

1. Elena Harman | May 26, 2010 at 7:51 pm

The Likert Scale should absolutely stay consistent throughout the survey. To change it on one question would be confusing to respondents. I agree that a 5-point scale is preferable over a 10-point scale, and if they really want more than 5 choices, 7-point Likert scales are the prefered method for pre-/post-testing and provide an opportinity for increase differentiation of responses.
Reply
2. Glenn | May 26, 2010 at 8:21 pm

Thanks Elena, I agree with you. The scale should stay consistent throughout the survey, it makes it clearer for the respondent – and allows for more comparable analysis. I think 7 points is the maximum. Most studies show that people find it difficult with scales greater than 7.
Glenn
Reply
3. mirely | May 27, 2010 at 7:50 am

definetely agree. consistency all the way. maybe benchpoint could provide the client with some references to help them understand why this is important…

glenn: enjoy your site very much
Reply
4. Richard Gaunt | May 27, 2010 at 1:21 pm

I am very grateful to everyone for taking the time to comment. Very helpful.
Reply
5. Glenn | May 30, 2010 at 7:03 pm

Thanks also for your comments on the blog!
Glenn
Reply
6. Marc Levasseur | June 7, 2010 at 10:38 pm

10 is more nuanced than 5. Of course you get more precise with the five possible answers but you give more room to respondents for a real appreciation with the 10 possibilities. As for the unity of methods…well hard to tell, this could be a challenge to calculate accurately with two different scales
Reply
7. Richard Gaunt | June 8, 2010 at 9:07 am

Ten or Five?
Marc has a point.
However it is quite complex to use words to describe 10 nuances of satisfaction/dissatisfaction.
And if you use a numerical scale, then you have to allow for individual bias. One person’s 7 may be another person’s 9.
At the reporting stage there would be an strong incentive (time, complexity) to take all those who answered 7 or more as “satisfied”, in which case you would have been better off with a 5-point Likert scale in the first place.
Does anyone have an example where a ten-point scale has actually made a positve differnence when reporting results, or implementing change as a result of a survey?
Reply