Type I Error vs Significance Level

A Type I error is a false alarm, the mistake of finding an effect that is not really there. Its long-run probability equals the significance level $\alpha$ you choose, so setting $\alpha = 0.05$ accepts a 5% chance of this error when the null hypothesis is actually true. For example, convicting an innocent defendant is a Type I error when the null hypothesis is innocence. Lowering $\alpha$ reduces Type I errors but makes Type II errors more likely, so the two trade off.

The significance level, written $\alpha$ (the Greek letter alpha), is the line you draw in advance for how much evidence counts as convincing. If the p-value is at or below $\alpha$ you reject the null hypothesis; if it is above, you fail to reject. For example, with $\alpha = 0.05$ you reject when the p-value is 0.05 or less. Because $\alpha$ is also the long-run rate of falsely rejecting a true null hypothesis, a smaller $\alpha$ such as 0.01 demands stronger evidence.