# So, what really is a p-value?

The p-value tells us the lowest level of significance at which we can reject the null hypothesis. What we mean by this is that if let’s say your testing level of significance is 5% or 0.05. If you are presented with a p-value lower than this, you can reject the null hypothesis.

Let me explain further using an example:

Now, a p-value is often referred to as the observed level of significance. It is compared to the level of significance used in the test to determine whether or not a given test or model is statistically significant.

Let’s take an example where you are testing an hypothesis that a given variable, let’s say, Education is significant in a given model. In other words, we are testing if this variable is statistically worthwhile or not in the model in question.

This hypothesis test could be presented as:

Now, let’s say that we are testing at the 5% level of significance.

If the p-value for the education variable is less than the level of significance of 0.05, you can reject the null hypothesis. For example, if your p-value presents as 0.01, therefore you can reject the null hypothesis (Ho) of insignificance and accept the alternate hypothesis of significance. Simple!

Note: Almost all statistical software packages will now provide p-values automatically. However, an element of caution is required here. These p-values are generally for two-tailed tests so therefore to be used in a one-tailed test, these p-values need to be divided by two.

Check out our blog post on one-tailed and two-tailed tests here.