Okay this blog is about your friendly proportion hypothesis test. First of all, let’s establish that your parameter of interest is a frequency count. You are concerned with how many times a particular event occurred during a time period. So let’s say you are conducting an experiment and you observe an event occurred 30 percent of the time (p=0.30). Then say you are sitting somewhere and your buddy Harry comes over with a scientific paper that claims the event was suppose to occur 20 percent of the time (p=0.20). And Harry just won’t let it go. He is really become a nuisance. He’s like listen you are wrong and this scientific paper is right!
So you say okay Harry let’s run a hypothesis test and I will show you that not only did I collect a quality sample- the event CAN occur 30 percent of the time! So you run a hypothesis test. Your null hypothesis is that Harry is a bone head but he’s right p=0.20. Your Alternative hypothesis is that Harry is still a bone head because you are right p>0.20. If you reject the null hypothesis, that means you collected plausible evidence that the event can occur greater than 0.20 (in this case p=0.30).If you fail to reject the null hypothesis then Harry is right for the sample you collected.
This is a general sense of how proportion hypothesis tests work but if you liked this blog there is more examples to follow…-Amy