Date: Fri, 21 Sep 2001 10:57:42 -0500 Carol Albright "SPSSX(r) Discussion" Carol Albright Re: Assigning cases to factors? text/plain; charset="iso-8859-1"

Hi, Peter

I'm not quite sure I follow what you're doing, but see if this helps -- You assign _variables_ to factors, not cases. Here's an example. Say you have a set of 25 survey questions measuring "quality of life" that you asked of college students. The questions are variables (columns) and each row is a case, all the data for one student.

You run a factor analysis using the 25 variables (survey questions) and learn that QOL is composed of 3 factors: How Nice My Professor's Are (Professor), Cafeteria Food (Food), and Relationships. You then compute FOR ALL CASES these new subscales (usually by adding up the values in the variables that make up a scale, or else taking the mean of the values). No case "belongs" to any factor.

Are you manually entering in the values? Don't -- use a formula (it's faster), such as:

COMPUTE food = q1 + q2 + q3 + q7 + q12.

or

COMPUTE food = mean.4(q1, q2, q3, q7, q12).

HTH, Carol

At 04:36 PM 9/21/01 +0200, you wrote: >Hi all > >When I do factor analysis I often want to know witch cases 'belongs' to >witch factors. >Right now Im doing a factor analysis and it results in 4 nicely defines >factors. >I now want to assign each case to one factor, so I do some computations >that finds the highest value in my 4 factor score variables. >If the highest value is in score variable one, I put 1 in my 'factor >belonging' variable and so on. (I know that a case can be defines by a >negative value too, but mostly I want the positive assignment). >My 'factor belonging' variable is devided into 4 approx. equal groups, and >I think this is a consequence of the distribution af the factor score >variable - mean of zero, std.div. of one and so on. >If I do a simple summation of the values in the variables I use in the >factor analysis - in groups identical to the result of factor analysis - I >get much greater diversity in the distributions of the four groups in the >'factor belonging' variable. > >I think really that my question is quit simple - how do I assign cases to a >single factor, without allways getting groups of equal size?? >Or have I compleatly misunderstood everything?? > >Thanks in advance > >Peter Løvgreen >Denmark > > ------------------------------------------------------------------------- Carol L. Albright, MS | E-Mail : syzygy@tc.umn.edu Albright Consulting | Phone : 651/699-7218 St. Paul, MN 55105 USA | Research data services http://www.tc.umn.edu/~syzygy (SPSS tips now on own pages) -------------------------------------------------------------------------

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