SPSS-creating groups question

TheNibbler

Hope this is the right forum for an SPSS question!

After a long (LONG) time away from SPSS I am trying to do some secondary analysis on a pretty large data set. As it is secondary analysis the data is not arranged in a nice straight forward way for me to run some quick calculations (the actual analysis is quite simple).

So my question is this: I will be looking at three groups of people but I need to distinguish between these three groups by their answers to two questions (both questions are entered as separate variables). IE. people who answer 'no' to Q1 are in group A and people who answer 'yes' are then further subdivided into group B and group C by answering 'yes' or 'no' to Q2.

I hope that is vaguely clear. Ideally I would create a new variable based on the above divisions. I could then go ahead and compare these three new groups on a number of different variables.

A colleague has suggested that SPSS Syntax might be the only way to do the above but as I have no idea how to use this I was hoping there would be a simpler solution?! I would imagine this is something that happens a lot but I don't think I've dealt with it before (at least not that I can remember).

Thanks!

Black Swan

TheNibbler wrote: »

Hope this is the right forum for an SPSS question!

You are welcome to post your SPSS research questions here.

Ostrom

What sort of analysis do you want to run?

You can just go ahead and create a new variable with 5 group codes (i.e. group 1, 2, 3, 4, 5) - this will allow you to split the sample, and you can use it as a categorical dummy if using multiple regression. The dialog procedure for this is [Transform -- Compute Variable], where you can specify the assignment process based on the criteria you provided above. You might be able to keep track of things better using syntax, however if this is a substantial piece of work, I would suggest investing your time getting to grips with a more intuitive and user-friendly syntax such as stata (selections and groupings like this are also much more straightforward, and you can manage your overall workflow much better).

If the structure of your research is more complicated, or there are repeat measures, you might need to code it different. Let us know what kind of analysis you want to run first.

TheNibbler

Thanks for your reply.

I'm in the early stages of working out exactly what we'll be doing with the data but it looks likely to be some descriptive statistics comparing the three groups followed by a multiple regression of each group. I'm not in my office at the moment and SPSS has decided to stop working on my laptop so I'll try your suggestion on Monday. That sounds like a logical solution and I'd prefer not to get bogged down figuring out syntax. It isn't a huge piece of work, something I'm helping a colleague to get published. I'll let you know how it goes on Monday.

Thanks

Ostrom

No problem, I only ask about the analysis because there is an intermediate step in setting up a categorical variable for regression in SPSS which is a bit annoying (it involves recoding the categorical variable into binary dummies).

Since you don't yet have the complete categorical variable, you can simply generate a binary for each group (i.e. create a variable for individuals answering yes to question 1 and so on), then enter these into the model.

TheNibbler

Ok, have had some success creating a new variable using the Transform-Compute variable facility. However I can't seem to figure out how to get it to separate into three groups.

I don't have much experience specifying these type of processes on any computer program so wasn't really sure how simple I could be. Essentially I need the following variable with 3 codes!:

1= (quitatt=0) 2= (quitatt=1 & pro3mcomp=1) 3 = (quitatt=1 & pro3mcomp=0)

I'm just not sure how to communicate this to SPSS in the correct terms. I originally tried to use the 'If' button but as I have two different variables I'm not sure how I can specify this using this method?

Edit: Just managed to figure it out, so simple in the end