SPSS help!

Discussion:

SPSS help!

(too old to reply)

Emily

2019-01-04 20:19:57 UTC

Hi,

I am writing a proposal for a simple psychology research project. Cross-sectional observational design.
2 IVs and 1 DV

My measures are:
AUDIT (Alcohol use disorder identification tool)
HSE stress indicator tool
DNRF (drinking norms rating form)

All result in ordinal data (I think?)

Which would be the best SPSS analysis to do on the data to find potential correlations? I have 1 group for my sample.

I also have a question at the beginning regarding whether participants have a parent in treatment for alcoholism. How can I compare this answer to participant AUDIT scores?

Thanks!!

Bruce Weaver

2019-01-04 21:16:23 UTC

Permalink

Post by Emily
Hi,
I am writing a proposal for a simple psychology research project. Cross-sectional observational design.
2 IVs and 1 DV
AUDIT (Alcohol use disorder identification tool)
HSE stress indicator tool
DNRF (drinking norms rating form)
All result in ordinal data (I think?)
Which would be the best SPSS analysis to do on the data to find potential correlations? I have 1 group for my sample.
I also have a question at the beginning regarding whether participants have a parent in treatment for alcoholism. How can I compare this answer to participant AUDIT scores?
Thanks!!

Hello Emily. Which of those variables is the outcome (dependent) variable, and how many levels does it have? Even it it really is ordinal, if the number of levels is quite large, it may be sensible (and defensible) to treat it as if it has interval scale properties--i.e., to use OLS regression.

On the other hand, if the DV is ordinal with a relatively small number of categories, you might consider using ordinal logistic regression (i.e., PLUM) instead. Here are some links to help with PLUM:

https://www.ibm.com/support/knowledgecenter/en/SSLVMB_25.0.0/statistics_reference_project_ddita/spss/advanced/syn_plum.html

http://www.restore.ac.uk/srme/www/fac/soc/wie/research-new/srme/modules/mod5/4/index.html

Two other comments not directly related to your question:

1. "SPSS help!" is not informative as a title for your post. See point #7 in the FAQ for the Statalist forum, for example (https://www.statalist.org/forums/help). David Marso made the same point (in rather colorful fashion) in point #7 of his tips for posting to the SPSSX discussion forum (http://spssx-discussion.1045642.n5.nabble.com/Tips-on-Posting-questions-to-the-SPSS-list-tp5715038.html).

2. Nowadays, the aforementioned SPSSX forum is far more active than this old usenet group, so you might consider posting there in future when you have questions. One easy way to join and follow that forum is via the Nabble archive. Click the "more options" link at the top of the page for info on how to join.

http://spssx-discussion.1045642.n5.nabble.com/

HTH.

Emily

2019-01-04 21:31:59 UTC

Permalink

Post by Bruce Weaver

Hello Emily. Which of those variables is the outcome (dependent) variable, and how many levels does it have? Even it it really is ordinal, if the number of levels is quite large, it may be sensible (and defensible) to treat it as if it has interval scale properties--i.e., to use OLS regression.
https://www.ibm.com/support/knowledgecenter/en/SSLVMB_25.0.0/statistics_reference_project_ddita/spss/advanced/syn_plum.html
http://www.restore.ac.uk/srme/www/fac/soc/wie/research-new/srme/modules/mod5/4/index.html
1. "SPSS help!" is not informative as a title for your post. See point #7 in the FAQ for the Statalist forum, for example (https://www.statalist.org/forums/help). David Marso made the same point (in rather colorful fashion) in point #7 of his tips for posting to the SPSSX discussion forum (http://spssx-discussion.1045642.n5.nabble.com/Tips-on-Posting-questions-to-the-SPSS-list-tp5715038.html).
2. Nowadays, the aforementioned SPSSX forum is far more active than this old usenet group, so you might consider posting there in future when you have questions. One easy way to join and follow that forum is via the Nabble archive. Click the "more options" link at the top of the page for info on how to join.
http://spssx-discussion.1045642.n5.nabble.com/
HTH.

Hi,

Thanks for your help. The DV would be AUDIT scores and they range from 0-4 if that is what you mean by levels.

I will attempt to change the post name and also check out the other group.

Bruce Weaver

2019-01-05 02:00:41 UTC

Permalink

Post by Emily

Post by Bruce Weaver

Hello Emily. Which of those variables is the outcome (dependent) variable, and how many levels does it have? Even it it really is ordinal, if the number of levels is quite large, it may be sensible (and defensible) to treat it as if it has interval scale properties--i.e., to use OLS regression.
https://www.ibm.com/support/knowledgecenter/en/SSLVMB_25.0.0/statistics_reference_project_ddita/spss/advanced/syn_plum.html
http://www.restore.ac.uk/srme/www/fac/soc/wie/research-new/srme/modules/mod5/4/index.html
1. "SPSS help!" is not informative as a title for your post. See point #7 in the FAQ for the Statalist forum, for example (https://www.statalist.org/forums/help). David Marso made the same point (in rather colorful fashion) in point #7 of his tips for posting to the SPSSX discussion forum (http://spssx-discussion.1045642.n5.nabble.com/Tips-on-Posting-questions-to-the-SPSS-list-tp5715038.html).
2. Nowadays, the aforementioned SPSSX forum is far more active than this old usenet group, so you might consider posting there in future when you have questions. One easy way to join and follow that forum is via the Nabble archive. Click the "more options" link at the top of the page for info on how to join.
http://spssx-discussion.1045642.n5.nabble.com/
HTH.

Hi,
Thanks for your help. The DV would be AUDIT scores and they range from 0-4 if that is what you mean by levels.
I will attempt to change the post name and also check out the other group.

I think ordinal logistic regression (via PLUM procedure) is more defensible, in that case. The restore.ac.uk link I posted earlier ought to be helpful.

Cheers,
Bruce