Quick Spss Question

billstraighten

Hi all

I have a problem,

I have 6 variables,
Eg.
A
B
C
AX
BX
CX
which are to be answered to, in ranking order of 1,2,3
However its possible that more than one can be ranked 1,2,3
Eg
A 1
B 3
C 2
AX
BX 1
CX 3

I'm quite rusty at spss, but i want to see if there is any correlation between picking A&B,B&C etc etc.
Does this mean i have 720 different variables

Any tips to get me started

2Scoops

Your post is confusing. You have six variables, right? If you just want to correlate them, I don't see what the problem is.

billstraighten

yeah, sorry i'm probly making it more confusing

Im just wondering what way should i input the data,

2Scoops

billstraighten wrote: »

Im just wondering what way should i input the data,

It looks like you want 6 columns: one for each of your variables. In each column, you will have the 'answers' to each variable [1,2,3 - ordinal or nominal]. Then go to Analyse -> Correlate -> Bivariate.

Is that what you were trying to do with the data?

billstraighten

Basically, i am using it for horse racing,
There are 6 factors
(ABCDEF)
that influence the results
1,2,3

Horses can have a combination of these factors
Ive covered 100 races

Ideally, i'd like to see what the combinations are
eg
A&D= 25 Wins
B&F=2 Wins

If i upload the .sav with the data put in, could someone tell me what i'm doing wrong

2Scoops

Interesting! I see what you're at now, but I wonder what the best analysis would be...

You could do straight counts which would be simple enough. Also, it seems like a good opportunity for some multiple regression. You can throw up the datasheet if you like and I might have time to go through it later in the week.

Also, don't forget the 7th factor 'G' - sekrit stuff the trainers know that the punters don't!

billstraighten

Haha, yeah i thought it would be a good topic to cover.

Thanks for the help 2Scoops

http://rapidshare.com/files/122710519/BANKERS_2.sav

There is the file, hope u can set me straight on a suitable layout, Im using spss 16 which is different to the one in college, maybe thats why :rolleyes:

2Scoops

Here are a few pointers that might help... but it might be a lot of work to rearrange to get some stuff.

In the current layout you can easily do proportions i.e. how many 1s, 2s, 3s in each factor. Pie charts should do it quickly. If there are remarkable differences then you'll see them straight away. Incidentally, reclassify 'nowhere' as 4, not 0, so it will be on the other end of the ordinal scale.

Following on from that, you could simply add up the wins for combinations of factors to see how that looks, but I think the number of horses that can be classified as being positive for multiple different factors will be small.

To get correlations between the factors and winning, you have to set it up very differently. The result (win, 2nd, 3rd etc) will have to get its own column, and each factor should get a nominal yes/no. So, for every winner, list yes/no for each factor; every 2nd place, every 3rd place, every nowhere. Basically, you need a row per horse - you'll end up with a thousand rows of data probably. :eek:

There might be a simpler way but I can't quite put my finger on it just now - I'll keep thinking about it for a while.