Hypothesis testing Flashcards
What is a hypothesis?
A hypothesis can be defined as a best guess or prediction of the outcome. One criterion is that any hypothesis must be falsifiable. There is usually two hypothesis the scientific H1 and Null H0. When testing a hypothesis, you are deciding on whether to reject the Null hypothesis or not.
What is usually the difference between H1 and H0?
The Null hypothesis (H0) is usually the more conservative/general of the two. Consider the example where you try to determine whether you are dealing with one or two populations. In such a scenario, the Null hypothesis will typically be that there is only one (the more general).
How do confidence intervals work?
A confidence interval uses a provided threshold for determining whether or not to reject the Null hypothesis. First we assume the Null hypothesis to be true, then we calculate the probability of the observations based on the Null hypothesis. Should the probability be outside our interval of confidence (the observations are too unlikely for the Null hypothesis to be true). Our definition of too unlikely resides on the selected level of confidence. If we use a 5 % significance level, we know that there is less than a 5 % chance of the observations made should the Null hypothesis be true.
What are the types of errors we can make in statistical decision making?
There are potential errors in statistical decision making:
- Type 1 error – falsely reject the null. This is controlled by the significance level.
- Type 2 error – fail to reject a false null.
We assume type 1 errors to be the worst, so these are the ones to be focused on. The type 1 error is also known as the active error.
What is a Significance Level?
The significance level (also called Type I error rate or the level of statistical significance) refers to the probability of rejecting a null hypothesis that is in fact true. This quantity ranges from zero (0.0) to one (1.0) and is typically denoted by the Greek letter alpha (a). The significance level is sometimes referred to as the probability of obtaining a result by chance alone. As this quantity represents an “error rate,” lower values are generally preferred.
