Steven Scott Bayesian writer and researcher has written many papers on the thoughts of Bayesian theory. He has lectured in many places around the US and even has his work on YouTube for you to follow. He is a firm believer and advocate for the Bayesian theory and works tirelessly for it to be understood and accepted worldwide. He believes it to be an extremely valid and effective way of testing hypotheses.  The purpose for statistics is to test a hypothesis to show whether it fulfils your thought or otherwise. In order to efficiently test for a hypothesis, you would have to repeat the test many times.

So, what is a hypothesis it is basically just an educated guess, an extra that you may wonder well if that works for a and b how about we try c. A hypothesis should always be testable.

If you want to write a hypothesis and you want to test it then you should be able to write a statement to sum it up. For example, if I do a then I expect b to happen. A good hypothesis should include an if and then, and also should include variables that are dependent and independent.  The hypothesis should be able to be tested either through experimentation, survey or another proven scientific methods. It should be based on research and have design criteria in some instances.

 Thus, hypothesis testing in statistics is a way that you can test a set of results when it isn’t apparently obvious what your testing shows. In other words, to find out if they are meaningful and whether or not they prove your hypothesis. The testing should prove that your results could not have happened by chance. It can be a confusing area and ways that can help are by:

  • Introducing a null hypotheses and stating what it is
  • Working out what test will help you prove or disprove your hypothesis

The null hypothesis is the accepted fact, the thing we already know that is the basis of our new testing. The thing we are looking to build upon. For example, there are eight planets in the solar system.

One sample is never enough but it is actually a theory. An example of this is that a professor thinks that his students are above average intelligence and he chooses a recent test to prove it. Already I can see that you are picking the holes out of this theory. It might be something they were particularly good at; there may have been a big incentive to do well on that test etc etc.

Another commonly used testing theory is Bayesian testing. This tries to repeat a previous test and get the exact same results, thus proving that both sets of results were accurate and correct and then using this to prove or disprove a hypothesis.

This therefore means that traditional testing methods will always require you to do many tests. Bayesian may not, depending on when you get your match.