Significance testing with Ctables & MRSets
Locked
 
|
Hello, I'm using the ctables function along with MRSETS to create custom tables of multiple response set questions. I'd like to test these for significance, however, so far what I've seen will test the entire variable against the entire variable For example Age vs Gender When what I want is to be able to determine whether or not specific aspects of age are significant. 18-25 yr old Males are significantly different from 18-25 yr old females, but otherwise there are not significant differences between males and females on age. The syntax I am using is below. This is to determine if there are significant differences in the desire for park services in different areas of the city. I'm volunteering on a park team for my city :)
*create the MRSet for facility. MRSETS /MDGROUP NAME=$icq10_rename Label='Which is most important facility? ' categorylabels=countedvalues variables=icq10a to icq10h value=1 /DISPLAY Name=[$icq10_rename]. *run ctables of areas by new facility set. CTABLES /VLABELS VARIABLES=!myvars area DISPLAY=LABEL /TABLE !myvars BY area [COUNT F40.0] + area [COLPCT.COUNT PCT40.1, TOTALS[COLPCT.COUNT PCT40.1]] /CATEGORIES VARIABLES=!myvars area ORDER=A KEY=VALUE EMPTY=INCLUDE TOTAL=YES POSITION=AFTER !ENDDEFINE. !XTABLES MYVARS=$icq10_rename. Thank you in advance for any assistance or guidance you can provide. Natalie --
"This was a lucky recollection-- it saved her from something like regret" |
Automatic reply: Significance testing with Ctables & MRSets
|
|
I am currently out of the office and will respond to your email, if necessary, on Thursday, January 24. If you need immediate assistance, please
call 812-856-5824.
Shimon Sarraf Center for Postsecondary Research Indiana University Bloomington |
Re: Significance testing with Ctables & MRSets
|
In reply to this post by Natalie
Natalie You can’t do significance tests on mult response sets. You need to turn each response into a dichotomy and test two-way (or more) crosstabs tables. As an example, if there’s a question with 10 possible responses coded 1 thru 10, are there 10 variables (M1 TO M10) or are there, say, only five (M1 TO M5) but using codes 1 - 10? An alternative approach would be to generate 10 dichotomous variables (one for each code) and use an appropriate measure of association. If you post more information about your data, I can give you more detailed advice and assistance. Have a look at the 3rd slide show in Old Dog, Old Tricks. SPSS usage in major surveys is a critical examination of SPSS usage in British Social Attitudes 1987 and European Social Survey 2002. Slides 15-25 show a multiple response question from the ESS about belonging to groups subject to perceived discrimination: slides 16 – 44 go through the steps needed in SPSS to analyse the data (actually the reverse of your own situation, but you can easily reverse the logic and the syntax). It looks as if you’ve used the GUI rather than syntax. Actually syntax is much simpler for what you want to do, so have a look at section 3.3 Multiple Response on my website. Hope this helps. John F Hall (Mr) [retired academic survey researcher] Email: [hidden email] Website: www.surveyresearch.weebly.com From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Natalie Spring Hello, *create the MRSet for facility.
Natalie -- |
Re: Significance testing with Ctables & MRSets
|
|
Natalie Glad it helped. Replies should go to the whole list so we can all see them, so that’s where I’ve posted this. I’ve found the GUI suitable for generating initial syntax for CTABLES, but I then trim it right down to basics and still get the same results as long as I accept the defaults. I’m still not quite used to it, but thanks to Jon (no H) Peck and Raynauld Levesque, I can now generate some fantastic tables to demonstrate elaboration (almost in the format I need) and I can get pretty close to them with the new subcommands for MEANS (trick of the trade). I’m a long way from generating tutorials to demonstrate these, but I’m considering posting one or two drafts on the site for feedback purposes. John From: Natalie Spring [mailto:[hidden email]] John, Thank you so much for this understandable explanation of what I should be doing and for providing resources. I've been using SPSS syntax for basic things for years, but have just started using MRSets and Ctables and clearly need to learn more in-depth how they are functioning. You response is incredibly helpful and actually makes sense as opposed to what I'd been googling. Natalie On Thu, Jan 24, 2013 at 4:40 AM, John F Hall <[hidden email]> wrote: Natalie You can’t do significance tests on mult response sets. You need to turn each response into a dichotomy and test two-way (or more) crosstabs tables. As an example, if there’s a question with 10 possible responses coded 1 thru 10, are there 10 variables (M1 TO M10) or are there, say, only five (M1 TO M5) but using codes 1 - 10? An alternative approach would be to generate 10 dichotomous variables (one for each code) and use an appropriate measure of association. If you post more information about your data, I can give you more detailed advice and assistance. Have a look at the 3rd slide show in Old Dog, Old Tricks. SPSS usage in major surveys is a critical examination of SPSS usage in British Social Attitudes 1987 and European Social Survey 2002. Slides 15-25 show a multiple response question from the ESS about belonging to groups subject to perceived discrimination: slides 16 – 44 go through the steps needed in SPSS to analyse the data (actually the reverse of your own situation, but you can easily reverse the logic and the syntax). It looks as if you’ve used the GUI rather than syntax. Actually syntax is much simpler for what you want to do, so have a look at section 3.3 Multiple Response on my website. Hope this helps. John F Hall (Mr) [retired academic survey researcher] Email: [hidden email] Website: www.surveyresearch.weebly.com From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Natalie Spring Hello, *create the MRSet for facility.
Natalie --
|
| Free forum by Nabble | Edit this page |