Replies to post #391437 on Anavex Life Sciences Corp (AVXL)

[Steady_T]

The P values are important. I would like to point out that the means/P values are not the only important aspect of the data.

The distribution of the responses is also very important. The number of subjects that improve over baseline will be a key factor in the approval decision as well as the number of subjects that exhibit no decline. Both of those populations are rare over the 48 week trial period. As Doc has pointed there will be a few in the placebo group that will improve and a few that halt decline but there will not be many.

If there is a population of "super responders" i.e. more than few, that will be a never before seen result.

[frrol]

The one-tail is probably appropriate, though the confidence levels and significance threshold we used might not be. We certainly didn't use the 'conservative' approach on this issue, but this is an ongoing debate in biostats.

[dayneyus]

amstocks82, on Investors Village, explains there is no need for Two tails as onetail showing significance Obviates the other in these 2 populations:

amstock82
..... "But first to the test. With books of the t-test, when you have a change in two populations over time that are being tested, a right-handed t-test or left-handed t-test is used. This is essentially a one-sided t-test.

In the case we have, the population that took A2-73 changed from the population that took the placebo.This is apparent in the data. We want to prove if the change in the population is unlikely to be random (p .05 or better) or it is statistically significant.

In t-tests, it is classically used for a few things. The one we are interested in is to show that one group's mean is different from another group's mean. Or the group that took Anavex 2-73 changed from the group that took the placebo.

A one sample t-test in this case is a statistical test where the critical area of a distribution is one-sided so that the alternative hypothesis is accepted if the tested population parameter is either greater than or less than the original population or placebo population, but not both. To put it another way, the placebo population should reflect the normal untreated Alzheimer's disease population that is being tested. To make sure, (double-blind) random selection of treatment patients versus placebo patients occurred. The test population is taking Anavex 2-73. The hypothesis being tested is that the overall population of patients taking 30 or 50 mg of A 2-73 will show a reduction in cognitive decline. Therefore, the population taking A 2-73 essentially on the graph moves/changes in one direction from the population taking the placebo. One of the classic equations for this is below. Note, there are some that are a bit different, but the following is good.".....

.