help with Likert analysis

[mcgeemr]

I am using 5 point Likert scale questions in questionnaires to measure people's attitudes towards new health technology. Can anyone help with the best way to numerically analyse the results?
Also - I have accidentally repeated one of the questions/statements.
How do I account for or deal with this in my results?
Thanks in advance, Marilyn.

[JohnM]

Marilyn,

One of my favorite topics!

Actually, the answer to your question depends on the purpose of the survey and the role of each question - if you could provide us with more specifics, we can answer your question more thoroughly.

I can provide some general guidelines, however:

There is a huge unresolved debate in the statistics world between those who feel that you must treat the scale as ordinal, and report the median, and those who feel that it's OK to treat the scale as interval and report the mean.

I distribute a 5-point Likert survey to my internal customers every year, and I report the mean - and I've never been led astray - I've been pretty successful in determining which issues are the most important to my customers.

So, personally, in most instances, I don't think it makes much difference. To be safe, you can report both the mean and median. In addition, market researchers often report the % of respondents who check off the "top box" (a score of 5 on a 5-point scale).

Another interesting approach with Likert scale analysis is to do a simple linear correlation between each question and a question which is designed to measure "overall" attitude. The questions that correlate highly with the "overall" question are what we call "drivers" - they have the most to do with the respondents' overall attitudes.

Hope this helps.

John

[mcgeemr]

Thanks to John for his quick reply.
Do you or anyone else have any comments about the other issue....
I included a repeated Likert statement unintentionally. Rather than simply remove one of them for analysis - I checked for agreement in responses between the two identical statements. Sure as fate - a few of the responses differed by at least one data point. How do I deal with this? Can I randonlytake one of the statements out for analysis or use the mean of the two responses for the offending duplicate statement?

Marilyn

[JohnM]

Sorry - I forgot to address this in my first post.

You'll need to average the results for the repeated question. So if on one of them a person checked off a 5 and on the other one they checked a 4, you'll need to count it as 4.5.