In hypothesis testing, the null hypothesis (H_{O)} states that no difference exists between the underlying populations from which two or more samples were taken. In other words, the null hypothesis states that no true change has taken place.

The alternative hypothesis (H_{A)} states that a true difference *does* exist between the underlying populations.

For example, when evaluating the effect of a process change on the average yield of the process, the null hypothesis would be

H_{O:} Yield_{old process }= Yield_{new process},

and the alternative hypothesis would be

HA_{:} Yield_{old process }< Yield_{new process}.