What role does the null hypothesis really play in the scientific process?

Posted by: chocolateraisons on: October 28, 2011

• In: Uncategorized
• 4 Comments

The null hypothesis has haunted my scientific investigations since Secondary school with no one ever really properly describing what it is or what is its purpose. When I was first introduced to the concept I was told that it was the opposite of the experimental hypothesis. So for this blog I am going to try and better understand what the null hypothesis is and what part it plays in research.

The Research methods textbook from last year described the null hypothesis as the “statement about the population or treatments being studied that says there is no change, no effect, no difference or no relationship.” The null states that there would be no change whereas the hypothesis aims to describe “a tentative description or explanation for the relationship between variables”. So in general terms the null hypothesis is the opposite of the experimental hypothesis. Where the hypothesis says that something will happen while the null hypothesis says nothing will.

Ok, that seems simple enough; one says that something will happen whereas one says that nothing will. However what is the point of having two hypotheses for the same theory?

Well one of the uses of the null hypothesis is that it helps to define what exactly the population would look like if nothing happened. So for example if you were doing research on the difference in mortality rates between those who when admitted to hospital suffering with depression; and were treated with anti-depressants, electroconvulsive therapy or neither (Avery, David and Winokur, George in 1976 Mortality in Depressed Patients Treated with Electroconvulsive Therapy and Antidepressants). So in this case the null hypothesis would state that there would be no difference between the three treatment conditions. There wouldn’t be any significant differences between the mortality rates for those who were treated with the electroconvulsive therapy, antidepressants or nothing.

The other advantage of using a null hypothesis is that it is much easier to completely reject than it is the reject the experimental hypothesis. This is because there is only one way the null hypothesis can be satisfied and that is if there is no difference. Whereas the experimental hypothesis is much more open and therefore could be satisfied through a number of ways. Using the previous example of the study on levels of mortality, the experimental hypothesis could be satisfied if those who received the antidepressants had a lower mortality rate than the other two treatment conditions, or if it had a higher mortality rate. Another way could be if the treatments of electroconvulsive therapy and antidepressant had a much higher mortality rate than the condition of not receiving either; and so on and so forth. So the null hypothesis sort of acts a bit like a shortcut, as it reduces the amount of testing that would be required to test all of the potential hypotheses that could be incorporated into the experimental hypothesis. In other words if there is any significant difference between the groups, the researcher can reject the null hypothesis straight of the bat; whereas if there is no difference between the treatment conditions, the experimental hypothesis can be rejected and the null accepted.

While the null hypothesis can seem as unnecessary and superfluous, it actually makes life and research much easier to understand and it helps to reduce the amount of time needed to form conclusions on pieces of research.