in order to have the same percentage of critical region on both sides of the distribution for a non-directional hypothesis as you would for a directional hypothesis (e.g. 5%) you would need to increase the alpha level. With 2 regions of 5% we now have an alpha of .10 instead of .05.

Why does this matter? This seems simple, why don't we just say if we conducting a 2 tailed test increase the alpha to .10?

Question

in order to have the same percentage of critical region on both sides of the distribution for a non-directional hypothesis as you would for a directional hypothesis (e.g. 5%) you would need to increase the alpha level. With 2 regions of 5% we now have an alpha of .10 instead of .05.

Why does this matter? This seems simple, why don't we just say if we conducting a 2 tailed test increase the alpha to .10?

kirbykirby · Answer

I'm not too sure if I understand your question exactly :S

kirbykirby · Answer

Are you saying that what they are saying is more "convoluted" than what needs to be said?

ganeshie8 · Answer

increasing the alpha increases the type 1 error rate

ganeshie8 · Answer

oh could be...  :)
if she is asking whats the problem with increasing alpha, then above is the problem ^^

kirbykirby · Answer

^right. But yeah I am not sure exactly what needs to be clarified o.O