about odds ratio again

[nurul]

my study is a cross sectional, and as i remember odds ratio , we calculate it in other studies, but in my litherture i found studies similar to mine presenting odds ratio(and i posted a question on how to do that and i got a great hints)

but is it right to calculate and present odds ratio

my second question when i select a variable to put in the row box to cross tab with my out come no and yes, all my variables are in catigories, so for those catigories that are 2, i got odds ratio, but for variable with more than 2 catigories, i dont get

what should i do

[Link]

For your first question, ORs are ok for cross sectional studies (in fact, cross sectional studies mainly use only ORs).

However, cross sectional studies themselves are not very strong evidence of whatever it is you're trying to show. This is because it is only a snap shot of what is presently going on and doesn't allow for any causal inferences.

For your second question, one possible thing you can do is collapse. If you cannot collapse, then I would suggest showing your data to someone who knows what they're doing.

[nurul]

can i do it this way:

since i have to calculate the odds only in contagious table.
if i have, say 4 categorical variable shown to be significantly associated with the out come variable by chi square .

1. i go and calculate the odds ratio only for these 4 variables not for every variable i have

2.and since each one have more than two catigories, each one i take to calulate the odds ratio for it , then i compare the odds between the first and second catigory and lefting the third as a reference.

(i found a study doing something like that , i dont know if i miss understood but i will just copy it to u so if i can do like this for my four significantly associated varibles please tell me:

""The odds ratios (OR) 2.615 (first year) and 0.969 (second year), when third year is considered as reference category also indicates statistically signi_ cant association. The odds of student having stress is higher in _ rst year, whereas the odds are decreasing in second year)

but even here, i have my outcome (no and yes)X first year students and what?

should i record the other catigories , second yr and third year in to other years then calculate the odds of yes among the first
or just calculate the odds like this:
first yr and third
then compare with
odds of second yr and third yr
(i.e. instead of recording , just using third catigory as reference one time to calulate the odds of the first year then with the third yr)

one last question , does it make difference that my outcome is in the spss as not having the out come and yes having the outcome (instead of yes and then no)


sorry, but see how much i am confused and strugling with only basic stat.

please be patient with me

[Link]

What you're suggesting is in essence logistic regression. You should look into that. Again, its very easy to mess up regression. I see people do it all the time. I would suggest sitting down with someone to help you do it correctly.