Skip to 0 minutes and 14 secondsSo then if you go back to one of the charts that we looked at and you put these results into perspective. You can see that we would check the but amour-propre at statistical test use. Yes and the for the primary outcome that could be to me yes or no, quite depending on whether which statistical test they used. And then know for some of them and that means again we would have to stop and then also they had yes for some of them because all a lot of the secondary outcomes were non significant.
Skip to 0 minutes and 44 secondsSo just some key things to remember with this is that for superiority studies when P is greater than point zero five it does not mean that it's equivalent or that it's similar even if the correct statistical tests were used and remember secondary outcomes that are non significant can't really tell you anything because you can't rule out that beta error and that it's difficult when they don't tell you exactly which tests were used. For each study to be able to apply that. So in this study I don't think the stats heard it.
Skip to 1 minute and 16 secondsIt didn't help it but I don't think it truly hurt the study results for this so just remember if they use the incorrect statistical test you cannot make any conclusion from the statistical results but that doesn't mean you can't look at the clinical results but then it's difficult to tell whether those differences were due to chance alone or not because everybody uses for the most part uses statistics or confidence intervals to determine the value of the results.
Skip to 1 minute and 42 secondsSo let's go through and do another example and maybe this time if you didn't stop the audio and do it on your own that I'd highly recommend that you print this slide and they take the example and fill it out before you listen to the next part of it. So it's easy to agree with me when I'm saying it but it's a little bit different to do it on your own or do it with a group together before I disclose the answers
Two keys on assessing alpha and beta errors
In this video, Prof. Mary Ferrill provides us some keys.
First, for superiority studies, when P is greater than 0.05, it does not mean that it is equivalent.
Besides, secondary outcomes that are not significant can not tell us anything because we can not rule out beta error.
Ultimately, print the blank table and fill it out if you did not do it on your own before.
After completing it, you can go to the next part.