I understand what it is but do not understand why it is needed.
I have read if you do not have one, and just have the alternative hypothesis, then there are so many possibility's that could cause the relationship between the variables. But if you collect data that proves the null hypothesis wrong then it doesn't necessarily suggest the alternative hypothesis is right and cant you come to the same conclusion as you would do without the null hypothesis (i.e it does/does not support your alternative hypothesis)? If you can prove the null hypothesis wrong with 95% certainty then why cant you prove the alternative hypothesis right/wrong with X amount of certainty?
Any help would be greatly appreciated, thanks!