About this lesson
A well written hypothesis contains two elements, the Null hypothesis and the Alternate hypothesis. Writing a clear hypothesis that can be quickly analyzed with a statistical test is a skill that will be illustrated and practiced in this lesson.
The hypothesis is a two part statement that forms the basis for the statistical test. This lesson explains and illustrates the format of the hypothesis.
When to use
The hypothesis must be written in order to determine what type of test is needed to accept or reject the Null hypothesis. Therefore, whenever you do a hypothesis test, you must first write the hypothesis.
It is no surprise that Hypothesis testing requires a hypothesis. When doing hypothesis testing for Lean Six Sigma projects, the hypothesis is a pair of statements about data associated with the process, product or problem being investigated. When well written, the two statements are opposites of each other. If one is true, the other cannot be true.
The two statements are always called the Null hypothesis and the Alternative hypothesis. Based upon how the statistical tests are performed, the Null hypothesis must always take the position that the factor being investigated has no impact on the process. Naturally, then the alternative hypothesis is that the factor does have an impact on the process. Or more broadly stated, the Null hypothesis describes the status quo, any changes are due to random chance. The Alternative hypothesis is often what we are trying to prove. If we think that a factor has an impact on performance, we want to show that effect statistically.
The characteristics of well-written Hypotheses are:
- They clearly identify the population being considered.
- They identify the dependent variable and independent variable(s).
- They state the type or direction of the effect – such as greater than or less than.
The Hypothesis test will then do a statistical assessment of the data and conclude to either reject the Null hypothesis or fail to reject the Null hypothesis. The focus of the Hypothesis test is too see if there is evidence of something different about the data. If that difference is found, then reject the Null. If there is no difference found, there is no reason to reject the Null.
Hints & tips
- Be clear about the population of data required and the measure you want to use. These are needed to determine which Hypothesis test should be used.
- An excellent way to provide tips is to show several examples:
- Example 1:
- Ho: There is no significant difference in test defects between the data from production line A and production line B.
- Ha: Production line A incidence of test defects are less than Production line B.
- Example 2:
- Ho: There is no statistically significant difference in repair rates between product made with supplier A material and supplier B material.
- Ha: There is a higher incidence of repair when using materials from supplier A rather than with materials from supplier B.
- Example 1:
Lesson notes are only available for subscribers.