cross tab using a variable with non-discrete categories?

Moderators: statman, Analyst Techy, andris, Fierce, GerineL, Smash

AsdaPriced
Posts: 2
Joined: Fri May 23, 2014 10:21 pm

cross tab using a variable with non-discrete categories?

Postby AsdaPriced » Fri May 23, 2014 10:25 pm

Hi there,

I am working on a project where I need to cross-tab survey responses by organizational type (i.e. cross tab 'how satisfied were you with this product?' (on a Likert scale) by 'what type of organisation do you represent?'.

This should be easy enough.

However, instead of having respondents categories themselves into discrete categories (e.g. I am a NGO OR a public sector employer OR a pressure group OR etc.), we gave them a multiple response question, and allowed them to categories themselves as up to two different organisational types (because some respondents might be both a 'consultancy firm' AND 'self employed').

So, I can't make a single non-dichotomous variable to do normal cross tabs, as the categories which would go into this 'organisational type variable' are not discrete.

I am guessing then that I will have cross tab using different filters instead of a single 'organisational type' variable (i.e. cross tab 'satisfaction level' by filter_NGO AND filter_public_sector_employer AND etc.), but I can't work out the syntax with which to do this.

Can anyone help? os: Windows, v: SPSS 20

(p.s. we would also like to do ANOVA tests on the resultant distributions, if this affects the action plan at all)

cheers for any help, and please feel free to redirect me to a different portal if I'm in the wrong place
RubenGeert
Posts: 100
Joined: Mon May 19, 2014 6:06 am

Re: cross tab using a variable with non-discrete categories?

Postby RubenGeert » Sat May 24, 2014 5:04 am

Dear AsdaPriced (not your real name, is it?),

I'll first make some guesses about your data: the first and second categories are two variables in your data, holding values 1 through n with value labels like "NGO", "IT company" ... The value labels are identical for both variables. If the first category has value 3 (self employed) then the second category may hold a different value or be empty. Values (1, 2) on category 1 and 2 has the same meaning as values (2, 1). The first and second categories are equally important.

If those guesses are correct, then you can merge the first and second categories into one variable. That's basically what's explained here http://www.spss-tutorials.com/combining ... s-in-spss/.

However, you should first take into accound that (1, 2) and (2, 1) should result in the same value. You can do so by creating two new variables, the first holding the lowest value of the categories and the second holding the other value or system missing if absent. Use APPLY DICTIONARY for transferring value labels from one variable to the other.

Let me know whether the guesses are correct and if all this makes any sense to you.

Kind regards,

Ruben Geert van den Berg
www.spss-tutorials.com
AsdaPriced
Posts: 2
Joined: Fri May 23, 2014 10:21 pm

Re: cross tab using a variable with non-discrete categories?

Postby AsdaPriced » Sat May 24, 2014 8:21 am

Thanks for this Ruben,

Your guesses would have made life simpler.

Unfortunately the people that designed the survey made around dichotomous variables ('are you an NGO? 0=No, 1=Yes', 'Are you a..., etc.').

Potentially making more problems, we are planning to merge these variables into around 6 different occupational groupings (i.e. all respondents that have said they are either a consultancy AND/OR a prevention firm AND/OR etc. would be grouped together): it is these groupings that we hope to cross tab with our other variables.

The issue however is if a respondent said they were both a consultancy AND a Small/Medium Enterprise (which we are placing in a seperate occupational grouping), their case would have to be considered twice in any analyses.

This is where my idea for using filters came into play (I'm in no way committed to this if anyone that think of an alternative).

Hope this has made it clearer, been a long week.

Cheers
RubenGeert
Posts: 100
Joined: Mon May 19, 2014 6:06 am

Re: cross tab using a variable with non-discrete categories?

Postby RubenGeert » Sat May 24, 2014 9:05 am

I still don't see any problem/difficulty but I need some time off so I'll probably have a look into this tomorrow, OK?
RubenGeert
Posts: 100
Joined: Mon May 19, 2014 6:06 am

Re: cross tab using a variable with non-discrete categories?

Postby RubenGeert » Mon May 26, 2014 6:35 am

OK, here's a suggestion:

-Combine the dichotomous variables into a single categorical variable. A really neat way for this is demonstrated here http://www.spss-tutorials.com/combine-d ... variables/.

-Perhaps also run some tests using each dichotomy as a separate independent variable and control for the number of tests you're carrying out.

Kind regards,

Ruben Geert van den Berg
www.spss-tutorials.com

Who is online

Users browsing this forum: No registered users and 1 guest

cron