What is Hypothesis Testing ?
Hypothesis testing is one of the statistical method used to confirm the effect that critical few inputs have on the outputs. Hypothesis testing must be used when the inputs are measured discretely. The outputs may be discrete or continuous. However the inputs must be discrete, if the inputs are continuous then correlation and regression testing may be used. The fundamentals of framing a hypothesis have been explained in this article:
The Logic behind the Null and the Alternate Hypothesis
Any hypothesis testing always has two hypothesis, the null and the alternate hypothesis. The null hypothesis testing shows no relation between the samples, whereas the alternate test accepts the existence of a relationship. Hypothesis testing therefore considers both the possibilities. It statistically reaches a decision as to which of the two hypothesis is valid.
The Null Hypothesis
The very name null signifies zero. The null hypothesis therefore implies no relationship in the variable parameters that are being measured. The null hypothesis states that there is no significant difference in the samples being measured.
For instance consider a sample of people being served at Branch A of a bank and customers being served at Branch B of the bank and service level is the parameter being measures. The null hypothesis will state that there is no statistically significant difference between the service levels at Branch A and Branch B.
Similarly the null hypothesis can be written for multiple branches. It can state that there is no statistically significant difference in the service levels of Branch A, Branch B, Branch C, Branch D and Branch E.
The Alternate Hypothesis
The alternate hypothesis by its definition is the one that is opposed to the null hypothesis. We never select the alternate hypothesis. When we reject the null hypothesis, the alternate hypothesis automatically gets selected. There are various types of hypothesis like:
Directional: A directional alternate hypothesis clearly states the type of relationship between the variables under question. For instance a directional alternate hypothesis will clearly state that the service level at Branch A is greater than service level at Branch B i.e. Branch A > Branch B. It could also use the less than sign.
Non-Directional: A non-directional hypothesis simply states that there is a statistically significant difference between the samples being measured. It does not tell us whether the Service Level of A is better or whether that of B is better. It merely tells us that they are different.
It is important to understand whether the alternate hypothesis should be written in the directional or non-directional form. This is because the statistical tests being used at the background change significantly.
Formulating the problem correctly maybe the most important role for the Six Sigma project person in the analyze phase. This is because there are tools which can automatically solve the problem, but that is only after they have been correctly formulated.