Question

we sometimes see in the literature terms like, "close to significance", "approaching significance", and "borderline significance", to refer to p-values that are close to 0.05. Some investigators do not think these terms are appropriate, i.e., something is significant or is not - thus, 0.049 is significant and 0.051 is not significant. How do you feel about this issue? No right or wrong answers, as long as you justify your response (and use citations).

Answer 1

P - value close to 0.049 and 0.051

As text book ( theory) tell that if p- value is grater than = 0.05 level of significancetaken we should reject Ho .But level of significance that means we may got wrong result 5 times out of 100.

We know that, p- value is nothing but probability. Which tell us about reject the null hypothesis or not. If p- value is very small ( less than 0.05) in such case we have more confident about rejecting null hypothesis and other hand if p- value 0.049 < 0.05 as theoretical rule says rejected null hypothesis.but in such case we may not more confident to reject null hypothesis. We know null hypothesis is nothing but testing something like mean , population variance or any thing related population using sample .

And p- value calculated using sample test statistic. Sample tastistics depends on random sample . So if we get p- value 0.049 for any test..and we take some more or less random sample once again it.may be come to less 0.049 or greater than 0.049 .

Data scientists or researchers or data analyst work on real life project if such situation comes in their project, so they not think it is appropriate results.

For example : bio- statistician work on to check which medicine is better for fever. In such case if p- value comes 0.049 ...means reject Ho i.e. not work same on fever.but in real life it is hard to say that medicine A is better than medicine B.because p- value is not very low..so he is not more confident to believe on this outcome.

We can take another one example apart from statistics..to shake of simplicity suppose you have weight 60.000 and your friend has 60.010 in such case in real life we say both have same weight because there may any other errors in measurements. But if we think mathematically your friend has more weight than you.

I hope you may understand better that why such issue in real life about p-value.