Hypothesis Testing is a statistical inference where a hypothesis is tested, as well as its null hypothesis, to determine whether to accept or reject the hypothesis based on a confidence value.

So there are two hypotheses. H1 and H0. We use the nature of a probability distribution to determine the likeliness that the samples we are testing are in the distribution. If the sample is very unlikely to have been drawn, less likely than the our confidence value, we reject the H1 and accept the H0. If the value is likely, more likely than our confidence value, we accept the H1 and reject H0.

In the best case scenario, we calculate the possibility that the probability of drawing using the z distribution, the normal distribution. We use the true mean(u), the sample mean(x), and the true standard deviation(s). (u-x)/s=z value. We next go to a Z table where find the appropriate z value and the probability. If that probability is above our confidence value, we reject. the confidence value is a percentage, which determines the area of the distribution where that percent of values is contained, meaning for a confidence value of 99%, 99% of the values will land within that area. So if our Z value is outside that area. We reject the hypothesis.

Back to DMAIIC Structure Other Types of Hypothesis Testing Kahn Academy’s Hypothesis Testing