🚀 go-pugleaf

You are putting the cart a long, long way before the horse.  Instead of asking 
what analysis to use, you must first ask yourself why you gathered the data in 
the first place.  What did you want to know?  Ask not what analysis to use, ask 
what you want to find out.  Then forget that there are fancy SPSS stat 
techniques and ask, from the basis of common sense, how your data might address 
what it is you might want to find out.  Now, still at the noncomputer level, 
looking only at frequencies, counts, means, and maybe a crosstab or two, see if 
it looks as if your data are telling you anything interesting about your 
question.  If not, the chances are that SPSS will not do much to save your 
bacon.  But if so, ah, NOW is the time to formulate specific questions to ask 
of your data, OK, now that you have a suggestion of what your data are saying, 
and a specific question to ask, perhaps SPSS will be able to help find the 
answer precisely and analytically.  But go to SPSS with your questions firmly 
in mind, and you probably will not have to come to this group for suggestions.

Ben Fairbank




In article <377483f5@news1.jps.net>, staltaro@jps.net says...
>
>Please help,
>
>I have finished collecting survey data from 330 psychologists across the
>country regarding their experiences with suicidal patients.  The data has
>been put into SPSS yet I am having a great deal of difficulty understanding
>how to analyze it.  I have a mediocre background in statistics, but I do not
>understand how best to approach the quantitative analysis of this data.    I
>have a series of 38 likert scale (ordinal) items which measure whether
>psychotherapists agree or disagree that they have had specific negative
>experiences with suicidal patients. In the same dataset are 10 demographic
>questions on gender, degree, specialty area, ethnicity, etc.. 330 cases have
>been entered in SPSS 6.1. How should I approach the quantitative analysis?
>Pearson shows many many significant associations of .2 - .4 but only two or
>three of .6 correlation. In addition, Chi-Square has found significance for
>every likert item. How do I examine reliability/validity? What stats should
>I use?   The 38 items are divided into categories based on the constructs
>they represent, there are 12 questions on narcissistic injury, 8 on
>aversion, 4 on malice, 3 on introjective identification, 1 on suppression, 1
>on reaction formation, and some others.
>
>Any suggestions would be dearly appreciated,
>
>
>staltaro@jps.net
>
>
>
>
>
>--
>
>*** Support Free Speech: Support Cryptography ***
>
>
>
>
>
>
>