[Afzal]

i am confuse again. we find out the p value 2.2% so,Since 2.2% is less than alpha (5%), we reject the null hypothesis. but i thought if it is less then alpha we not going to reject.

[Ezra Halleck]

The null hypothesis is always what you are trying to disprove. The pvalue measures what is the chance of getting something as extreme as you found assuming the null hypothesis. The lower the pvalue, the stronger the evidence that the null hypothesis is wrong.

[Alphatron]

What’s confusing is knowing what your null hypothesis is. Sometimes a problem will state that something has been discovered by someone and then that person will attempt to prove that his discovery is correct. For example, the toothpaste company problem where the company itself tried to prove that it’s findings we’re accurate.

They were trying to prove the they were right, yet still, the null hypothesis would be exactly what their findings were. The null hypothesis would be that their toothpaste reduces cavities.

[Ezra Halleck]

I apologize for orginally stating the toothpaste problem wrong. I at first was thinking that I would take the point of view (POV) of a regulator. Then I switched midstream to the POV of the company.

From the regulator’s POV, she wants to prove that the toothpaste is junk, that it is not an improvement, that the advertisement is false, so the null hypothesis is the opposite, which is that it actually works.

From the company’s POV, it wants to prove that the toothpaste works, that it is an improvement, that the advertisement is true, so the null hypothesis is the opposite, which is that it actually is junk.