Date: Wed, 21 Feb 2007 14:00:29 -0500
Reply-To: Art@DrKendall.org
Sender: "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From: Art Kendall <Art@DRKENDALL.ORG>
Organization: Social Research Consultants
Subject: Re: Survey Scale Analysis
In-Reply-To: <200702210413.l1KMc3be020484@mailgw.cc.uga.edu>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
*First *proofread all of your data.
This is critical in any analysis.
There are ways to enter data twice and compare them but they require
that users be more experienced. *Ironically, although this is one the
first tasks for raw beginners, it has not yet been put on the DATA menu.*
Also, you will revise your syntax as you develop your understanding of
SPSS and of your data, always exit menus via <paste>.
If you have items that were reverse worded to counterbalance social
desirability, response effects, etc. then
recode var1 to var24 (else= copy) into newvar1 to newvar24.
* this is an instance where it is ok to recode items into themselves
since you can retrace your steps.
recode var3, var4, ..(1=5)(2=4)(3=3)(4=2)(5=1)(else=copy).
reliability . . .
The example syntax is not tested. substitute variable names etc that
fit your particular situation.
You can check the syntax for a procedure by highlighting it and clicking
the syntax icon.
JC wrote:
> I am new to SPSS and need help working with survey data. I administered a
> survey to 60 students. There are three separate scales on the survey. Each
> scale has 7-8 questions with five possible responses (almost never, not very
> often, etc.). Only one answer is allowed per question. I entered the scale
> questions as separate variables (q12, q15, etc) in the variable view and
> gave values for each participant's response to each question (values = 1-5).
>
In the future, it helps respondents treat extent items as interval if
you anchor "none or almost none" with a zero.
> I need to do two things:
>
> 1) Calculate each participant's (60 separate cases) mean score for each of
> the three scales. The overall mean for each scale can range from 1-5.
>
After the reliability run has verified the the items have been properly
reflected, keyed to the correct scales, and that no items should be dropped.
compute score1 = mean.5 (newvar5, newvar7, ......).
compute score2 = mean.5 (newvar1, newvar11 . . .).
compute score3 = mean.5(newvar . . .
formats score1 to score3 (f4.1).
> 2) I should end up with a profile for each participant that looks something
> like this:
>
> Participant #1 Mean score for scale #1(7 questions)= 3.5; Mean score
> for scale #2 (8 questions) = 4.0; Mean score for scale #3 (7 questions)= 3.1
>
print records = 5 table
/ 'participant #' ID
/ 'Mean score for scale #1 (7 questions) =' score1
/ 'Mean score for scale #2 (8 questions) =' score2
/ 'Mean score for scale #3 (7 questions) =' score3
/.
> 3) Then, for each scale I need an overall mean and std deviation for all
> participants, so three overall means and three std devs. Maybe this is
> called a group mean and std dev for all cases/participants on each separate
> scale (?)
>
descriptives vars= score1 to score3/statistics = all.
> If this is a compute function then please be very specific about the
> necessary syntax. I already tried the compute option and the default
> function, for example for one scaleâEUR" MEAN(var1,var2, var3, var4, var5, var6,
> var7)/7 âEUR"and this didn't work. I tried to name that new variable BES (S for
> scale) and was still unable to compute 60 separate means for the 7 questions
> on that scale. The new variable appeared useless.
>
You would have had mean/7. not sum/7.
> Sorry this is long but I want to be clear. I am very frustrated at not being
> able to do this simple thing on SPSS when it is a snap on EXCEL.
>
> Regards,
>
> JC
>
>
>
