If you want to compare the mean values, you can use a one sample ttest on there differences compared 0.
I am comparing two blood glucose meters. I am using a blood sample from the same finger, at the same time using two different blood glucose meters.
I have 125 paired test results; 125 test results from one meter and 125 test results from the 2nd meter.
Do I use “T-Test for 2 Independent Means” because the blood sample is tested by two independent blood glucose meters?
or
Do I use “T Test for 2 Dependent Means” because I am using the same blood sample?
Thank you!
If you want to compare the mean values, you can use a one sample ttest on there differences compared 0.
Stop cowardice, ban guns!
(Which is equivalent to a paired t-test “T Test for 2 Dependent Means”). It makes sense to treat the observations as paired because each pair of measurements is from the same blood glucose measurement.
Note: The t-test is just investigating whether the means of the two methods are the same. You may also want to check how closely related the observations are from the two meters. Regression or correlation could tell you about this/
Is one considered the gold standard? Also look to examine their dispersion (variance) .
Stop cowardice, ban guns!
Tweet |