A simple way to know P-value
and what it says about the data
“In statistical hypothesis testing, the p- value or the probability value is the probability of obtaining test results at least as extreme as the results actually observed during the test, assuming the null hypothesis is correct.”
After reading this, I was like what does this even mean?
To better understand this let’s take an example
Let’s say there is a pizza store near your house and they claim to deliver it in 30 minutes or less. But as a hardcore pizza lover, you believe it’s more than 30 minutes they take it to deliver because sometimes you found your pizza a bit cold than usual.
This left us with two options
- Null Hypothesis(H0)- What pizza store says delivery time is 30 min or less.
- The alternate hypothesis(H1)- what you believe the delivery time is more than 30 min.
You might be thinking we are talking about the pizza and suddenly what this weird terminology( null/alternate hypothesis)comes along.
Before I overcomplicate the things further, I will give you a bit of background about them.
- p-value helps us to determine the significance of your results when we do the hypothesis testing in statistics.
Hypothesis testing is used to test the validity of the claim that is made about the population. In our case is pizza delivery time which is a max of 30 min that is claimed by the pizza store. And this claim(delivery time max 30 min)we have put on trial to test is called Null Hypothesis.
And opposite of is this is the alternative hypothesis that what you believe is true(delivery time more than 30 min)if Null Hypothesis is concluded to false.
The evidence we found in the trial (about the delivery time) uses a p-value to determine the strength of the evidence( what we can determine from data about the population(i.e overall delivery time of pizza).
The p-value always between 0 and 1 and gives us the below interpretation
- p-value≤ 0.05 indicates strong evidence against the null hypothesis, so you reject the null hypothesis( means pizza takes more than 30 min to deliver)
- p-value > 0.05 indicates we have weak evidence against the null hypothesis, so we cannot reject the null hypothesis.
- p-values very close to the cutoff (0.05) are considered to be marginal (could go either way). Leave it up to to users draw conclusions about p values just provide the p-value to them.
Now come back to our pizza we randomly select some delivery time of pizza and run a hypothesis test. You found that the p-value is 0.001 which is lesser than 0.05. So the claim of pizza is being delivered in 30 min or less is false.
Normally we reject the null hypothesis when the p-value is less than 0.05. So we conclude that pizza place is fooling you around and their delivery time is more than 30 minutes on average.
And the only thing that you would like to know that what will pizza store going to do about this delayed delivery. (Of course, you could be wrong by having sampled an unusually high number of late pizza deliveries just by chance.)
Reference-
