Choose the correct test - did I do right?

#1
Hi there.

First of all: Sorry for bad english.

I got 2 problems concerning my survey. I try to figure out if there’s a correlation between the number of injuries and the number of training.
n = 477

1) I used a modified CMDQ-test (excerpt ist to be seen at 1st attachment) – so there were 23 body parts participants had to give information about.

I grouped them as shown in 2nd attachment: (I hope it's okay in german, it's all about the numbers)

Then I did Spearman-Test (because data wasn’t normally distributed) to know whether my hypothesis can be confirmed or not. Turned out that my hypothesis can’t be confirmed - for each of the body parts. All results were between p = .173 and .952

A) Did I do the proper test?
B) And is it necessary to group the data? If I don't do it I got different results - e.g. several body parts got p < .05



2) Whether if I did wrong or not - following step should be next: I would like to get an overall result not only for the body parts. But how do I move on?

May I’m thinking way too complicated… Do I have to sum up the data of all the Spearman results and calculate an average?


SPSS / statistics seems to be all Greek to me. :oops:

It would be awesome if you could help me!
 

Karabiner

TS Contributor
#2
Spearman is appropriate. Use the orginal (ungrouped) training frequencies; categorizing such data destroys statistical information.
You cannot simply calculate the mean of the 23 Spearman coefficients, but to report the the median should be o.k. here.