Critical values are fundamental components of statistical hypothesis testing that serve as decision thresholds for accepting or rejecting null hypotheses. These values mark the boundaries of rejection regions in probability distributions and play a crucial role in determining statistical significance. This article explores what critical values are, how they're calculated for different distributions, and how to interpret them in various testing scenarios.
A critical value is a point (or points) on the scale of the test statistic beyond which we reject the null hypothesis. In other words, it's the cutoff value that defines the boundary between the region where we accept the null hypothesis and the region where we reject it in favor of the alternative hypothesis.
Critical values are determined by:
The significance level (α) represents the probability of rejecting the null hypothesis when it is actually true (Type I error). Common values include 0.05 (5%), 0.01 (1%), and 0.10 (10%).
The relationship between α and critical values depends on the test type:
The Z-distribution is used when the population standard deviation is known or when the sample size is large enough (n ≥ 30).
For a Z-distribution:
Two-tailed test:
Left-tailed test:
Right-tailed test:
Common Z critical values:
The t-distribution is used when the population standard deviation is unknown and the sample size is small (typically n < 30).
For a t-distribution with df degrees of freedom:
Two-tailed test:
Left-tailed test:
Right-tailed test:
The degrees of freedom (df) for a single sample t-test is n-1, where n is the sample size.
The chi-square distribution is commonly used for testing goodness-of-fit, independence in contingency tables, and variances.
For a chi-square distribution with df degrees of freedom:
Two-tailed test:
Left-tailed test:
Right-tailed test:
The degrees of freedom depend on the specific test:
The F-distribution is used primarily in analysis of variance (ANOVA) and for comparing variances of two populations.
For an F-distribution with df₁ (numerator) and df₂ (denominator) degrees of freedom:
Two-tailed test:
Left-tailed test:
Right-tailed test:
For an ANOVA test:
The rejection region represents the set of values for the test statistic that leads to rejecting the null hypothesis. The critical value defines the boundary of this region.
For different test types:
Two-tailed test:
Left-tailed test:
Right-tailed test:
Suppose we want to test a hypothesis with α = 0.05 using a two-tailed Z-test.
The critical values are:
The rejection region is: z < -1.96 or z > 1.96, or |z| > 1.96
For a two-tailed t-test with df = 15 and α = 0.05:
The critical values are:
The rejection region is: t < -2.131 or t > 2.131, or |t| > 2.131
For a right-tailed chi-square test with df = 5 and α = 0.01:
The critical value is:
The rejection region is: χ² > 15.086
When conducting hypothesis tests, you compare your calculated test statistic to the critical value:
For example, if you calculate a Z-statistic of 2.5 for a two-tailed test with α = 0.05, since 2.5 > 1.96, you would reject the null hypothesis.
While critical values define rejection regions based on predetermined significance levels, p-values represent the probability of observing a test statistic as extreme as the one calculated, assuming the null hypothesis is true.
The relationship between them:
Critical values can be found using:
Critical values have diverse applications across multiple disciplines:
Critical values are used to construct confidence intervals. For example, a 95% confidence interval for a mean using the Z-distribution has boundaries μ ± 1.96 × (σ/√n), where 1.96 is the critical value.
Changing the significance level changes the critical value. A smaller α (e.g., 0.01 instead of 0.05) makes the critical value more extreme, requiring stronger evidence to reject the null hypothesis.
Yes, for distributions that include negative values (like Z and t distributions), critical values can be negative, especially for left-tailed and two-tailed tests.
As degrees of freedom increase, the t-distribution approaches the standard normal distribution. Similarly, critical values for chi-square and F-distributions change with degrees of freedom.
Two-tailed tests distribute the significance level between both tails of the distribution, while one-tailed tests place all of it in one tail. For example, with α = 0.05, a two-tailed Z-test has critical values at ±1.96, while a one-tailed test has a critical value at ±1.645.
Parametric tests (like t-tests) use critical values from specific distributions assuming certain conditions about the data. Non-parametric tests (like the Wilcoxon signed-rank test) may use different methods to determine critical values, often based on rank statistics rather than the data itself.