First, I'd strongly encourage you to delete the "undecided" response category because it is generally not monotonic in its relationship with the other categories.
Next, Rasch analysis can be used to both evaluate the psychometric properties of your scale and create a true interval summary score that can be used for further analysis.
For an introduction to the Rasch model, see Bond and Fox, Applying the Rasch Model (2nd ed).
Scott R Millis, PhD, MEd, ABPP (CN,CL,RP), CStat
Professor & Director of Research
Dept of Physical Medicine & Rehabilitation
Wayne State University School of Medicine
261 Mack Blvd
Detroit, MI 48201
Email: [log in to unmask]
--- On Fri, 1/16/09, marina garcia <[log in to unmask]> wrote:
> From: marina garcia <[log in to unmask]>
> Subject: likert scales
> To: [log in to unmask]
> Date: Friday, January 16, 2009, 11:39 AM
> hi allstat,
> i have a problem with 5-point likert scales on how to
> analyze them statistically. i have a survey question that
> asks the respondent to choose an answer from 1=strongly
> agree, 2=agree, 3 = undecided, 4=disagree, 5=strongly
> disagree about a certain feeling...
> 1. is it ok to get the mean (or the weighted mean) of the
> responses of the respondents to the survey question and use
> this to describe how they feel towards that question?
> 2. what test should be used in order to determine if there
> is a significant difference in the way the respondents feel
> towards the question if the respondents are grouped into,
> say, 3 groups? in short, is there a significant difference
> among the 3 groups in the way they respond to the question?
> can the one-way anova be used or is kruskall-wallis better?
> the internet provides with many answers but i'd like to
> ask you for your opinions...kindly help me...please..:-)
> marina garcia
> Surf the Web in a faster, safer and easier way:
> Download Opera 9 at http://www.opera.com
> Powered by Outblaze