choosing the right analysis with spss 12.0

[tdk]

dear friends
i am a new user of SPSS and fan of statistics.As you can understand, it really surprized me to find out that there is a site like yours, with so much interesting topics in statistics and so much help.It is the first time to post. So here is my problem
"i have two groups of patients with urine stones in the kidney and the upper ureter to be treated with ESWL. The first group is positioned in the supine position,whereas the second group is positioned in the lateral position.The aim of my study is to see if the second group can be treated more efficient than the first group and even more to see if the outcome of the ESWL in each and between groups depends on the location of the stones,i.e.kidney or ureter."
I would be grateful if you could help me to choose the right statistical analysis with SPSS.
with kind regards
TDK

 

[JohnM]

Do you have an equal number of patients in each combination (position x kidney stone location)? In other words, in each of the 4 combinations:

(1) supine / kidney
(2) supine / ureter
(3) lateral / kidney
(4) lateral / ureter

do you have an equal number of patients?

 

[tdk]

thank you for your reply
i have two groups:60 patients in the supine position and 60 patients in the lateral position.Of the 60 patients of the first group, 44 had kidney stone and 16 had ureteral stone.Of the 60 patients of the second group,44 had kidney stone and 16 had ureteral stone too.

 

[JohnM]

You could use factorial ANOVA (2 factors --> patient position and stone location) with "ESWL outcome" as the dependent variable, or you could use logisitc regression, but this depends on what type of variable the "ESWL outcome" is.

Is it a numerical result or is it categorical (successful / not successful)?

 

[tdk]

Well,the outcome of ESWL is defined as "successful/not successfull". Would you advise me to use factorial ANOVA or logistic regression?
P.S. A friend of mine suggested to use chisquare test or independent t-test for the total of groups (position/result)and then again chisquare or independent t-test for each group (location/result).Of note,he is not a statistical expert!!What do you think?

 

[JohnM]

Actually, chi-square tests would be easy to do if you only had one independent variable - you could have made a 2-way table showing position or location vs ESWL result.

- this can be done with 2 independent variables, but it's a bit more complicated (called "log linear analysis") - and is basically analyzing two 2-way tables:

(1) patient position vs ESWL result (stone location = kidney)
(2) patient position vs ESWL result (stone location = ureter)

and examining any change in the success rate for position, and also looking at how it changes, if at all, based on stone location (i.e., interaction between patient position and stone location).

Here are links to references on analyzing 2x2x2 or 3-way tables:
http://www.socialresearchmethods.net/tutorial/Cho/222table.htm

http://faculty.ucr.edu/~hanneman/soc271/ctab3.html

and here's a great reference on how to do this type of analysis in SPSS:
http://www2.chass.ncsu.edu/garson/pa765/logit.htm

 

[tdk]

Dear sir
thank you for your valuable advices. Yet,considering the fact that i am not an expert in statistics,would you suggest me to use log linear or logistic regression analysis?What would you suggest it would be more easy to do with SPSS 12.0?-
P.S.1 Of note, my version of SPSS does not include logistic regression.
P.S.2 In spite of that though, do you think that logistic regression is the most suitable analysis for my study?
Thank you in advance..!!

 

[JohnM]

Since you don't have logistic regression, I would go with log linear. They're very similar.