LISTSERV at the University of Georgia
Menubar Imagemap
Home Browse Manage Request Manuals Register
Previous messageNext messagePrevious in topicNext in topicPrevious by same authorNext by same authorPrevious page (May 2003)Back to main SPSSX-L pageJoin or leave SPSSX-L (or change settings)ReplyPost a new messageSearchProportional fontNon-proportional font
Date:   Fri, 2 May 2003 08:24:45 -0400
Reply-To:   Art@DrKendall.org
Sender:   "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:   "Arthur J. Kendall" <Art@DrKendall.org>
Organization:   Social Research Consultants
Subject:   Re: Multiple regression Newbie
Comments:   To: SBeasley@burswood.com.au
Content-Type:   text/plain; charset=us-ascii; format=flowed

Some reactions: The purpose of having items is usually to sum them in a psychometric scale. How many constructs (scales) are the 20 items designed to measure? Are some for the independent (predictor, X) variable(s) and some for the dependent (criterion, Y) variable?

If you only have 12 cases, are the items perhaps, in some repeated measures set up: pre-post, from dyads, from couples, multiple time points, etc.?

Are you sure that the items are ordinal? Or are they Likert items which can usually be used to create a Likert scale? Likert items themselves are not very discrepant from interval level and the summative scale is clearly not.

With only 12 cases, unless the phenomenon is extremely robust, the main outcome of a regression run will be to familiarize yourself with how to run a regression, with what kind of information appears on an output, and how the information is laid out in the printout. Quantitative interpretation of the results approaches being silly with 12 cases and no repeated type structure in the data. Be sure to also graph your data.

You said you are a "newbie", so I presume this is just an exercise. There are more advanced approaches if you had more data, but get the basics down before going to the esoteric. It would be unusual for a summative psychometric scale to have merely ordinal response scales. Responding to whether a factor analysis would be more appropriate if you were to have more data would depend on a fuller description of your data and researchable questions.

Hope this helps.

Art Art@DrKendall.org Social Research Consultants University Park, MD USA (301) 864-5570

Simon Beasley wrote: > Hello all, > > being new to the SPSS application I am wondering if anyone can help me with > this question. I need to run a multiple regression on a small data set (12 > observations) 20 items each observation. Unfortunately I can't run a > factor analysis on such a small data set. It is ordinal scales as a part > of a psychometric test for attributional data. Any thoughts to help a > newbie would be appreciated > > > Regards > > > Simon Beasley > Business Analyst - Burswood CRM Project > Tel: (61-8) 9362 7742 Fax : (61-8) 9470 1789 > Mob. 0422 320 522 > Burswood International Resort Casino > > > > > ------------------------------------------------------------------------ > > ****************************************************************************************************** > > Burswood International Resort Casino > Winner of the 2002 Sir David Brand Award for Tourism Excellence. > > ****************************************************************************************************** > > Privacy and Confidentiality Notice > The information contained herein and any attachment is intended solely for the named > recipients. It may contain privileged confidential information. If you are not an intended > recipient, please delete the message and any attachments and notify the sender of > mis-delivery: Any use or disclosure of the contents of either is unauthorised and may be > unlawful. All liability for viruses is excluded to the fullest extent permitted by law. >


Back to: Top of message | Previous page | Main SPSSX-L page