Anyone good with SPSS

[ktulu]

I know this is a strange place to ask but is anyone good with SPSS. I am trying to run some stats for my thesis and I cant remember how to compare the periods like I want to. I have about 10 time periods as one variable and I want to go through and do Chi-square comparing two of the time periods at time with the other variables. If anyone can remind me how to do that or give me direction to a good website that lays it out that would be great.

Thanks in advance,
Ktulu

 

[jimscott]

It's been awhile since I've learned and used the chi squared distribution but a quick Google yields this:

http://en.wikipedia.org/wiki/Chi-square_distribution

http://www.statisticssolutions.com/Chi_square_test.htm

I remember vaguely something about the observed minus the expected divided by two and something about degrees of freedom and a maybe one minus alpha over two. I'd have to re-study it. I once made predictions about how I thought the NFL records would turn out one year (expected) and waited for reality (observed).

 

[herenorthere]

I wish I had it running on my home computer because I'd probably be able to step you through it. Are your data in an appropriate format?

 

[ktulu]

Thanks both of you,

Jimscott - thanks for the links but I am trying to get SPSS to do what I want, I already know that Chi-Square is appropriate and what it tests, but thanks again for trying to help.

herenorthere - I have already run Chi-square on the data as a whole but I believe its obscuring the the differences between the proto-Nasca and the Nasca (Nazca) periods, the Nasca and the Middle Horizon time (Wari or Huari) and the Middle Horizon and the Late Intermediate period. (These periods probably mean nothing to you don't worry) These periods are being compared with another variable that is essentially a yes/no question. The thing I can do is separate out these periods and compare them to each other to measure if there is a significant difference in the presence of the second variable.

 

[herenorthere]

It sounds like a job for the "select cases" function, which you get to by clicking on data. That will let you limit the analysis to two periods. When you get to the select cases window, you'll get a list of variables and highlight the name for your period variable and on the button which limits it to specific cases. I can't remember what it says, but it should be fairly obvious. That'll open a box that lets you say the variable = 1 or 2, assuming Proto-Nasca is coded as 1 and Nasca is coded as 2. If you get it right, you'll see a diagonal line through the row number box of all the rows where the period is something other than a 1 or a 2. That way, when you run the analysis with the period being one of the variables, those are the only two being considered. I hope that's what you're trying to do. I've logged a number of hours on SPSS but have to re-acquaint myself with it every time. So these directions might be a little off, but they'll hopefully get you heading in the right direction.