Even vs odd Likert scales

Adrian Liem adrianliem at telus.net
Fri Dec 16 23:12:19 MST 2005


I spent some time in 3rd year psych looking at the effectiveness of Likert
scales. It's been a while, but if memory serves me correctly...

One study (and unfortunately I can't find it nor remember the references)
found a "statistically significant" pattern where certain respondents tended
to disregard the extreme ends of a scale. The conclusion they reached here
was that if you are aiming to gather data covering a 5-point spread, you
might consider using a 7-point scale under the expectation that a
significant number of your respondents will disregard 1 and 7, leaving you
with 2-3-4-5-6 as your five points to measure your data.

I can see strengths and weaknesses with building a survey under this sort of
assumption. I can also see the pros and cons with even-vs-odd Likert scales.

There are so many variables that can affect the validity and reliability of
data gathered through surveys. Whether you go with a 5-point scale, a
7-point, 3, 4 or 6, I think an important point to take away (pardon the pun)
is that the context under which you're administering the survey makes a
difference. And so long as you (a) appreciate the strengths and weaknesses
of whatever scale you've chosen and (b) explicitly consider how your survey
design may be affecting your results, what will probably matter just as much
in the end (if not more), is how the data is interpreted into information.

Adrian Liem
Vancouver, BC


-----Original Message-----
From: bounce-stcusesig_l-222577 at lists.stc.org
[mailto:bounce-stcusesig_l-222577 at lists.stc.org] On Behalf Of Kit Brown
Sent: Friday, December 16, 2005 4:44 PM
To: STC Usability SIG discussions
Subject: [stcusesig_l] RE: Even vs odd Likert scales

HI Caroline--

Thanks for the article. Interesting...I hadn't quite considered it that way.
I wonder if this might be a cultural difference between Brits and Americans,
or perhaps the degree of investment in the results of the particular survey
(e.g., asking veterinarians about their professional development needs vs a
more general how do you like this product)...When I've done surveys with an
even number of questions, I've received much better responses to the even
scale than to the odd. As with anything, the degree of granularity/nuance
needed might dictate which scale gets used...

Regards,
Kit Brown
Principal
Comgenesis, LLC
PO Box 140092
Boise, ID  83714
+1 208.343.8706
kit.brown at comgenesis.com
 
"Practice random acts of kindness and senseless beauty"


-----Original Message-----
From: bounce-stcusesig_l-202598 at lists.stc.org
[mailto:bounce-stcusesig_l-202598 at lists.stc.org] On Behalf Of Caroline
Jarrett
Sent: Friday, December 16, 2005 1:19 PM
To: STC Usability SIG discussions
Subject: [stcusesig_l] RE: Question For the Day: What are the different
types of d...


Sorry to say, I very much disagree with you on this point. If the 
respondent's opinion is that they want to sit on the fence, then 
that's what their opinion is. I can't agree that making them decide 
about their opinion is a good idea. In fact, I go so far as to say 
that I think eliminating the middle point is somewhat disrespectful. 
The effect of doing so, in practice, is that you create two middle 
points as respondents who persist will just pick one or other of them.

For a longer article on this point, see:
http://www.usabilitynews.com/news/article1269.asp


Caroline Jarrett




---
You are currently subscribed to stcusesig_l as: adrianliem at telus.net
To unsubscribe send a blank email to leave-stcusesig_l-222577R at lists.stc.org

-- 
No virus found in this incoming message.
Checked by AVG Free Edition.
Version: 7.1.371 / Virus Database: 267.13.13/198 - Release Date: 12/12/2005
 

-- 
No virus found in this outgoing message.
Checked by AVG Free Edition.
Version: 7.1.371 / Virus Database: 267.13.13/198 - Release Date: 12/12/2005
 





More information about the stcusesig_l mailing list