Re: How not to design a Likert Scale?

William - interesting discussion. I'm in agreement with Max about  
treating ratings and survey scales differently.

I like putting labels on the scales for survey questions to ensure  
users are working with a consistent mental model. In William's  
example, it seems that the need for authorship got in the way of the  
goal of creating an easily understandable scale that ins't open to  
interpretation so it can yield unequivocal results. The beauty of  
"standard" likert scale divisions is that they are hard to  
misunderstand.

Whenever I depart from the norms, I make sure to use parallel wording  
pairs for positive and negative (Always/Never, etc.,) and  always  
preflight with a test group/ go back and ask if any of any scale  
seemed confusing or was open to interpretation.

With ratings, endpoints only work because the user can easily grasp  
where s/he is on the continuum.

All this has worked well for me for quite a while. Maybe I'm just  
lucky. It's doable to create friendly or audience-related (read  
"hip") questions AND still have the necessary rigor. It just takes  
paying attention and getting the ego of authorship out of the way.

My $0.02.

Marilyn Tahl

+ + + + + + + + + + + + + + + + + +
Marilyn Tahl

c 650.868.6601  | o 530.333.2968
[email protected]

"Empathy is not just about stepping into another's shoes. First you  
must remove your own shoes."  - Indian proverb




On Jan 23, 2009, at 7:34 AM, Max Harper wrote:

> An interesting discussion!
>
> One point I would add is that we might treat "ratings scales" (as in
> the movie examples above) and "survey scales" differently.

    --------------------------------------------------------------
    Tip of the Day: Quote only what you need from earlier postings
     CHI-WEB: www.sigchi.org/web POSTINGS: mailto:[email protected]
              MODERATORS: mailto:[email protected]
       SUBSCRIPTION CHANGES & FAQ:  www.sigchi.org/web/faq.html
    --------------------------------------------------------------