Advice on type of test to use

#1
Hi guys.

I have performed a study looking a 2 groups of people and how a particular blood test may be related to a disease.
My two groups are
1) people diagnosed with disease
2) people without disease ie. controls.

I have results that show people with disease have a noticeable "spike" in the blood level.

I would like to determine whether this is truly "significant" or whether this is by chance.

I was planning on using an unpaired two tailed t test.
Is this the correct type of test?

I have compared the means of the two groups, and they are not significant. I have also compared the "peaks"/"spike"values of the disease against the mean of the control. This is significant.

Any suggestions/help would be greatly appreciated!
:yup:
 
#2
Its not quite clear what the raw data look like here. Do you have a time series? i.e. do you have a set of data points for each person? Or just one data point per person. If there is just one point per person, the data are interval type (continuous), and the data are fairly bell shaped, then an unpaired t-test for means will work.

I think you will need to describe a little more what you mean by "spike" values. An excel table might be helpful.

~Matt