FW: spss syntax help

[John F Hall]

Thought I’d copied this to the list: looks like Art got it as well

 

 

John F Hall (Mr)

[Retired academic survey researcher]

 

Email:   [hidden email] 

Website: www.surveyresearch.weebly.com

Start page:  www.surveyresearch.weebly.com/spss-without-tears.html

  

 

 

 

 

 

From: John F Hall [mailto:[hidden email]]
Sent: 15 April 2013 07:24
To: 'William Fleming'; '[hidden email]'
Subject: RE: spss syntax help

 

Michael

 

I’m assuming that your data set has 90 questions * 5 dichotomised binary responses coded 0,1.  This makes 450 binary variables for each respondent.  Each question will have a string of 5 digits, only one of which will be 1, the other 4 being 0.

 

Respondent ID may or not be present in the data: data matrix consists of N rows x 450 columns.  I’m assuming a 90 x 450 matrix with no respondent ID.  This can be obtained later from the row number.

 

What software are you using?

 

I’m working blind here but if I had the data set in front of me I’d be doing something like this in SPSS:

 

data list

file ‘external file’

records 1

/1 V1 to v90 (1-450).

 

This creates 90 5-digit variables, only one of which will be 1, the other 4 being 0.  The trick is in the data transformation, and I think you need something like.

 

recode v1 to v90 (10000 = 1)(01000 =2)(00100=3)(00010=4)(00001=5)

  Into Q1 to Q90.

 

Alternatively

 

recode v1 to v90 (10000 = 1)(1000 =2)(100=3)(10=4)(1=5)

  Into Q1 to Q90.

 

. . and get an initial idea of what the variables look like with

 

freq q1 to q90.

 

If your questions are arranged in blocks related to a single topic each, I’d start computing scores within blocks: eg for a block of 5 related questions in v81 to v85

 

compute score1 = sum.5 (v81 to v85).

 

You can quick summaries of single response categories using MULT RESPONSE in dichotomous mode:

 

mult resp groups (q1 to q90 (5))

/freq groups.

 

Will yield a table with a row for each question indicating how many people endorsed 5.

 

After that you might be looking to use analysis based on correlations.

 

Once I’m happy with the data I’d add variable and value labels.

 

If your data are in an Excel file, SPSS can import data direct (including column headers, or you can simply copy/paste the matrix contents direct into the SPSS Data Editor.

 

If your data are not binary but coded 1 – 5, where each value constitutes a separate variable, it gets more complicated (and tedious).

 

Hope this helps.  If you send me (an extract from) your data matrix, in confidence,  I could be more specific.

 

 

John F Hall (Mr)

[Retired academic survey researcher]

 

Email:   [hidden email] 

Website: www.surveyresearch.weebly.com

Start page:  www.surveyresearch.weebly.com/spss-without-tears.html

  

 

 

 

 

 

 

From: Survey Research Methods Section of the ASA [[hidden email]] On Behalf Of William Fleming
Sent: 15 April 2013 05:47
To: [hidden email]
Subject: spss syntax help

 

Good Evening All,

It is late...and I just discovered a problem with a data set that has been passed to me that I'm hoping some clearer, less tired folks might help me with resolving.

I know it should be pretty simple, but it's not coming to me at this hour.

I have a data set with approximately 90 questions that were set up on  5 point likert scale.  The data that was down loaded from survey monkey dichotomized each response for each question. Thus for each of the 90+ questions I have 5 questions that have been dichotomized. I want to convert these into the likert scale it was suppose to be.

Can anyone point me in the right direction?

Thanks in advance.

Michael
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ To subscribe/unsubscribe SRMSNet: http://listserv.umd.edu/cgi-bin/wa?A0=srmsnet&D=0&F=&H=0&O=T&S=&T=1

SRMS website: http://www.amstat.org/sections/srms/ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~