Statistics?

Question

RE: statistics - Significance level/ Null Hypothesis/P-value,?

1)
Assume a significance level were set to .05 and the p-value for an upper-tailed test was .051. Given how close the results were to getting a “reject Ho” decision, what would you suggest the researcher do to improve his/her chances of rejecting the null hypothesis while still maintaining ethical standards?

will it descrease/increase sample size/ Significance level? or Rewrite the null hypothesis?

2)
If a null hypothesis is rejected at the .01 significance level, then it would also be rejected at the .05 significance level.
is this true or false

Answer 1

1) The textbooks all say you set the CI (level of significance) before you do the hypo...and stick to it. Changing it after the fact, to not reject the null, would raise real skeptics of your conculsions and damage your credibility. I would personally not do it.

2) True. An alpha (level of sig.) marks the areas under the upper/lower tails of the probability distribution. So when alpha = .01 you are saying "I'll reject my null when my test result falls into this area under the curve because the probability it fell there randomly is very slim (.01)." Or another way to say it "Something must have caused it; so it was not a random event." And the null is about random outcomes.

So when you set alpha = .05, you are making it easier (more likely) that a random event would fall in that area. Or less likely something actually caused it to fall there. But, if your result fell into the .01 area of the distribution curve, it would also fall into the .05 area; so it would have been rejected at that level as well.