In desperate need of statistics help... data reduction

[towsonu2003]

I am using SPSS and trying to write my thesis...

I did a factor analysis on some of my variables and I need to recode the factors into a new variable. The variables in the factors are 5-point likert scales. My teacher wanted me to just add each variable in a factor.

So, if factor1 consists of var1, var2, var3, and var4; according to the teacher, it would look like:
factor1=var1+var2+var3+var4
So the max for factor1 will be 20 (4 variables x 5th point in each Likert scaled var) and the minimum will be 4 (4 variables x 1st point in each Likert scaled var)

The question is: Is this the way to go? Shouldn't I add coefficients for each variable somehow? If so, how do you determine what those coefficients will be??

So, what I am thinking of is:
factor1=(coef1 x var1)+(coef2 x var2)+(coef3 x var3)+(coef4 x var4)

Thanks in advance for all help :)

[towsonu2003]

sorry :( to bump this so early, but I'm really desperate (by tomorrow morning)... :cry:

[towsonu2003]

factor analysis>save>use regression (new variable will appear at the end of the variable list), use primary component analysis option in factor analysis (factors now called components)

[xequence]

It's all greek to me.

[towsonu2003]

It's all greek to me.
same here...

I found a very nice book on this (thanks to "SPSS Wiki" clues): "advanced and multivariate statistical methods", which is kinda greek->english dictionary for this weird stuff.

[xequence]

same here...

I found a very nice book on this (thanks to "SPSS Wiki" clues): "advanced and multivariate statistical methods", which is kinda greek->english dictionary for this weird stuff.


Shouldent your teacher have thought you how to do this before making you do it? If not, theyre odd.

[towsonu2003]

the teacher is my advisor... I jumped into this as my thesis, thinking that my undergrad statistics education would be enough (no good courses on statistics - quantitative in grad courseload though!), but it seems it wasn't.

the solution was simple though: read more... I'm used to reading by now thanks to linux ;)

[briancurtin]

the solution was simple though: read more... I'm used to reading by now thanks to linux ;)
converting to all linux has helped me in this regard too. ive learned how to dig through documentation and how to search for what i need much, much better. i didnt really think about that until recently.

[towsonu2003]

converting to all linux has helped me in this regard too. ive learned how to dig through documentation and how to search for what i need much, much better. i didnt really think about that until recently.
we need a new policy: force students to learn linux regardless of graduate area of study ;) that'll teach them (us) how to work with literature reviews and other documentation...

[WolfJay_83]

It's quite easy to use the compute function to create a new variable. Then you can simply click each variable and sum them up. This will create a new variable with the above properties.

I cant remember where on the menu it is, as I'm on my ubuntu laptop right now. I'm curious though, are you running spss on linux somehow?

[towsonu2003]

It's quite easy to use the compute function to create a new variable. Then you can simply click each variable and sum them up. This will create a new variable with the above properties.
It wasn't giving me the variable I wanted as I didn't know what coefficients to use... I could just add each variable to each other, but that didn't look right... the "save" option with regression under factor analysis did the trick. what it basicall does is create a new variable with mean=0 and stD=1, using some coefficients unknown to me, from the components it extracts. I still don't know how it gets the coefficients though...

I didn't get any feedback from my advisor, so I'm not sure whether this will be okay for her... we'll see... I may have to just add them up as she wanted initially, if she insists... it seems advisors are gods in the thesis world...

I cant remember where on the menu it is, as I'm on my ubuntu laptop right now. Transform>Compute I'm curious though, are you running spss on linux somehow?
nope... in Windows (department computer and own computer). but I end up reinstalling windows from time to time bc. for some reason, the version I use at home in spss craps itself out if I want it to do too much computing: make it come up with a couple of graphics and plots, u end up having to reinstall it. I wish my undergrad people thought me how to use R instead of SPSS!! why the hell do you teach ppl a proprietary thing for which they have to pay big bucks to use???? academics is weird! I'll learn R once this is over...

[diafanos]

It's all greek to me.

even though I am Greek...I can't help you either...:lol:

[euler_fan]

. . . I wish my undergrad people thought me how to use R instead of SPSS!! why the hell do you teach ppl a proprietary thing for which they have to pay big bucks to use???? academics is weird! I'll learn R once this is over...

This might help in your quest to learn R coming from an SPSS background.

R for SAS and SPSS Users (http://oit.utk.edu/scc/RforSAS&SPSSusers.pdf)