Null hypothesis



Statistics is data collection in order to later organize, analyse, interpret and also present them in a specific manner that gives inside look in the problem and probable solutions in the area that is being studied. It can be used in many spheres from science to social and industrial fields. One of the most prominent hypotheses that is used very often in statistics in the null hypothesis, because in this discipline in many cases the null hypothesis is assumed true until evidence proves otherwise.

The null hypothesis in general is a statement or default positions that suggests that between two specific measures phenomena there is no relationships. Therefore with the help of statistics researcher need to determine that there is a relationship between two phenomena in order to  disprove the null hypothesis.

The null hypothesis also know as ad denoted as H0 is used in two very different statistical approaches. In the first approach called significant testing that was patented by Roland Fisher the null hypothesis can be disproved on the basis of data that is under the assumption. But in the second approach – hypothesis testing – that was introduced by Jerzy Neryman and Egon Pearson, the alternative hypothesis is put against the null hypothesis and then the truth is distinguished basing on the data, keeping in mind error rates.

The null hypothesis was established in 1925 by one Roland Fisher, although even before then there were talk about similar concepts in the statistical research and testing community. Fisher announced the null hypothesis and made it the main way to analyse almost all of experimental science back then. Later in 1933 Neyman and Pearson came of with an improvement to Fishers test, but later that became an alternative to the Fishers test rather than enhancement. And so these two ways of using the null hypothesis in statistics became to be and are still used all over the world.

One of the fields this null hypothesis is used is testing the significance of differences in treatment and control groups. At the beginning of the test it is assumed that there is no difference between the control and the experimental groups or any other two variables for that matter. For example with this method you can test and see if there is any difference in two groups, even if you only have one random sample of test scores from men and other – from women. Then first you need to assume that both test scores are the same and that can be showed with equation H0: μ1=μ2, where H0 is the null hypothesis but the μ1 and μ2 are the result of easy groups test scores.  Now all you need to do is to overthrow this statement and you will have overthrown the null hypothesis.