Re: T Test Question
Hi Jason,
What you would want to know is whether or not different test methods give statistically identical means and analytical variances. Apply Fisher’s F-test to verify that analytical variances are statistically identical. If they are, then apply Student’s t-test to paired test results. If analytical variances differ significantly, then apply Student’s t-test with a pooled variance. Of course, you may test for associative dependence between paired test results for the different test methods.
Jan W Merks ■
Re: T Test Question
Thank you so much for the reply. I have one more question about interpreting the results. For the Student T-Test, we want the values to be insignificant(i.e. >.05) if we want to say there is not difference between the two methods correct? If this is the case, is there some way that we could relate the correlation values to being highly significat with a different test? Thanks for all the help again!
jason ■
Testing Different Test Methods
If you took two sets of replicate test portions from a single test sample, and applied a different test method to each set, you should apply Student’s t-test with a pooled variance. If you took duplicate test portions from a set of test samples, and applied different test methods to each of the duplicate test portions, you should apply Student’s t-test to paired differences between duplicate test results. Only in the latter case does it make some sense to apply correlation-regression analysis. ■
Re: T Test Question
Thank you for the help. We do have duplicate test portions from a set of test sample so I think the paired test is the way to go.
Is there a way to describe the p-value in terms of being the means not being different but having a signficant p-value? ■
Stats For Paired Data
So the t-test is applied to paired data. If you wish, test for associative dependence between paired data by different test methods. The r-value ought to be significant if the variance of the set of test samples is higher than the analytical variances between duplicates. The intercept of the regression line is bound to be significant if the mean difference between test methods is statistically significant. ■
T Test question
Hello-
I have a question about T Testing with non-similar measurement methods. We have paired samples that were ran through 2 different testing method(Affy Array and QTPCR). We are trying to validate the results using the 2 methods to see if the basic trend of results (increasing or decreasing levels) is the same between the 2 tests. We used a Pearsons correlation to check if the 2 sets of data are positively correlated and we want to apply a p-value to the tests. I guess my first question is are we even taking the right approach with the T Test, and Second, are there any other method/problems to consider when comparing such results. Thanks for all help provided and I truely am sorry if this has been answered over and over again.
jason ■