How to compare paired data sets include the variation on the samples
Jun 16, 2020 6:58 AM(362 views)
Feeling i meeting the wall with comparing these two methods I'm testing.
I have 2 methods (1 and 2) both analysed on the same sample. For both methods triplicates have been measured for each sample. So now I want to compare method 1 vs 2 including the know variation to see if statistically significant?
What is this study, an analytical method comparison? An equivalence study for instrument replacement? Is one of the methods a reference? What do you mean by, "So now I want to compare method 1 vs 2 including the know variation to see if statistically significant?"
Your data is pre-summarized, so you you could use Bivariate platform with the mean and then select Fit Orthogonal to compare the null hypothesis of an intercept = 0 and a slope = 1. You could use Matched Pairs platform if you are looking for a difference in the mean.
I would recommend that you include the individual replicate measurements instead of summarizing the data into the mean and standard deviation. You could use a Fit Least Squares after stacking the measurement columns. The measurement in the Y role and then enter Sample as a random effect.
Hard to say much more given we know practically nothing about your study, your intentions, or your needs.