在进行假设检验之前，我们必须明确规定决策标准，即显著性水平（significance level），也称α水平，其实质是一个特定的概率。在心理学假设中我们一般使用的α水平是α=5%或者α=1%，这些极端值构成的区域被称为临界区域（critical regions），当样本均值落在其中时，我们认为数据与虚无假设不一，因此拒绝虚无假设。
选择决策标准时同样存在检验的方向性的问题，如果检验是没有方向性的，概率分布曲线的左右两端都属于临界区域，因此我们也将其称为双尾假设（two-tailed test），这时如果我们事先规定显著性水平α=5%，那每一段临界区域包含的范围实际上只是总样本均值数量的2.5%。
如果检验是有方向性的，临界区域只分布在概率分布曲线的一段，我们将其称为单尾检验（one-tailed test）。
Before conducting a hypothesis test, we must clearly specify the decision criteria, i.e., the significance level, also known as the alpha level, whose essence is a specific probability. In psychological hypotheses we generally use an alpha level of α = 5% or α = 1%, and the region formed by these extreme values is known as the critical region.When the sample mean falls in it, we consider the data to be inconsistent with the null hypothesis and therefore reject the null hypothesis.
The issue of testing directions also exists in choosing the decision standard, if it is not in a certain direction, the two sides of the distribution curve are both critical regions, so it is also called two-tailed test. If the alpha level is α = 5%, then each critical region only includes 2.5% of the mean of population.
If the test is directional, the critical regions only exists in one side of the distribution curve, it is called one-tailed test.