Comparing four groups

#1
Hello,

The study includes a total of 293 patients and they are divided into four equal groups by the leukocyte count. The data was collected from the registries. The distribution is non-gaussian. I quantified how many patients have diabetes in each group and now I would like to see if the difference between the number of diabetic patients in each group is statistically significant. What test should I perform?

Thank you!
 

Karabiner

TS Contributor
#2
Why did you think you have to create 4 groups, instead of using the
leukocyte count as is? You could compare mean leucocyte count between
those with vs. without diabetes, using for example a Welch-corrected t-test.

You can additionally describe the rates of those with within the 4 artificial
groups anyway, but for a statistiscal test your categorization seemingly is a
waste of statistical information, and it lacks generalizability (if you created
4 groups according to the leucocyte count distribution in the sample, instead
of using generally accepted categories).

With kind regards

Karabiner
 
#3
Why did you think you have to create 4 groups, instead of using the
leukocyte count as is? You could compare mean leucocyte count between
those with vs. without diabetes, using for example a Welch-corrected t-test.

You can additionally describe the rates of those with within the 4 artificial
groups anyway, but for a statistiscal test your categorization seemingly is a
waste of statistical information, and it lacks generalizability (if you created
4 groups according to the leucocyte count distribution in the sample, instead
of using generally accepted categories).

With kind regards

Karabiner
Thank you for your response!