Multiple t-tests in Excel adjusting for Type 1 error
HTML-код
- Опубликовано: 9 сен 2024
- This video describes how to run multiple t-tests in Excel looking for differences between more than two groups and adjusting for Type 1 errors. Normally, a t-test is limited to testing for differences between two groups. When looking for differences between more than two groups an ANOVA test is used. However, when running an ANOVA in Excel you are unable to run a post hoc test to determine which group is different from which. An ANOVA only tells you whether or not there is a difference between any of the groups being tested. Running multiple t-tests using the Bonferroni correction to adjust for Type 1 error is a way to work around this limitation when using Excel for testing multiple groups.
Thank u so much sir..
AMAZING SIR! Thank You!
GOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOD JOB!!!!!
Thanks a B
I don't understand why the amount of stores compared should have any effect on whether store A is different from store B. So if store A and B were the only two with data, then there is a significant difference. But if store C and D are considered also, then all of a sudden the difference between A and B is insignificant? Surely this doesn't follow logically, right? If drug A is more effective than drug B and could save lives, then why would things all of a sudden flip if scientists decided to consider drug C and D? I think the problem here is more scientific confidence in the p-value and the all or nothing true/false value many put behind it, making it seem like a categorical rather than continuous value. Regardless of this question of whether p-value should be used, it still doesn't follow logically that a drug could be considered effective if compared to one other drug, but considered ineffective if compared to 3 other drugs. Sorry for the long comment, especially as I know you didn't come up with this type of testing you are simply showing how to implement it in Excel, I've just been a good bit confused on this type of testing and the reasoning behind it, as to me there seems to be more problems with using it than not using it.
Why use the two tail vs the one tail?
Thank Mr. Olson. But I think it is not clear why the p value was divided by 6. If you only have 4 groups. thanks
It has to do with number of possible combinations; He has 4 groups, but there are 6 possible t-tests here. For example, you can run a test comparing 1:2, 1:3, 1:4, 2:3, 2:4, and 3:4. These are all the possible combinations.
I watched this video as I am comparing 6 groups to each other. There are 15 possible t-tests to run, making my new p-value 0.008 as the cutoff for significance.
Cheers
Thanks, Cortez, that makes sense. It wasn't very clear to me all the combination factor.
Thank you Cortez
thanks for this video, please can you tell me how can we know about the test numbers? here he has 6 can you explain the numbers of the test.
What if my ANOVA analysis has a P value of 0.022867 (suggesting there is statistical significance between at least 2 of my 4 groups), but when performing t-tests (with the p-value set at 0.00833) none of the 6 combinations show any with statistical significant difference. Is this possible? Or a calculation error?
Hi Beth, Your question is a little beyond my level of knowledge, but this article may help: www.ncbi.nlm.nih.gov/pmc/articles/PMC3916511/
From my understanding, I think your findings are not in error and are possible. The issue is probably with interpreting the p value. However, you should seek someone with a little more experience than me. Good luck!
Thanks for this information. Sir can you tell me where is calculated value in this description. I am confused when I have written description. So pls clear this point. Thanks
Hi Manne, My apologies for the late reply.
I'm not sure what you mean by calculated value. The video begins describing some of the calculations used.
I hope this helps.
@@ChristianHigherEd Calculated value means result of t test calculation
How do you do it if you have 3 data slots instead of 4?
If we want to do an F-test for checking variances for multiple data sets, do we also divide Alpha on the number of comparisons?
I'm sorry, but I don't fully understand your question. Please expand a little.