Normal-Curve Tail Probabilities and the Single-Sample Z Test

The way we calculate this probability depends on whether our research question is directional:

One-Tailed Probability

A one-tailed test is used for directional questions—for example, testing if a new teaching method makes scores specifically "greater than" the current average. For a positive $z$, the one-tailed probability is expressed as $P(Z \geq z)$.

Two-Tailed Probability

A two-tailed test is used when we want to know if a value simply "differs from" the mean, without specifying a direction (higher or lower). This measures the combined area in both the upper and lower tails. For a symmetric positive $z$, the two-tailed probability is $P(|Z| \geq z)$, which is effectively double the one-tailed area.

Exam Tip: As the $z$-score increases, the tail probability decreases significantly. This reflects the fact that extreme values (those many standard deviations away from the mean) are increasingly rare in a normal distribution.

The Single-Sample Z Test

While z-scores help us locate individual values, the Single-Sample Z Test allows us to perform statistical inference on sample means. We use this test to determine if a sample mean ($\bar{X}$) differs significantly from a known population mean ($\mu_0$).

This test is specifically applicable when:

1. The population mean ($\mu_0$) is known.
2. The population standard deviation ($\sigma$) is known.
3. Observations are independent.
4. The sampling distribution is approximately normal (often guaranteed by the Central Limit Theorem).

Standard Error vs. Standard Deviation

It is vital to distinguish between the standard deviation and the Standard Error (SE). While the standard deviation describes the variation among individual values in a population, the standard error describes the variation among sample means.

The Standard Error is calculated as: $SE = \frac{\sigma}{\sqrt{n}}$ where $n$ is the sample size. As the sample size increases, the standard error decreases, meaning larger samples produce more stable and reliable estimates of the mean.

Step-by-Step Implementation

To conduct a Single-Sample Z Test, follow this structured procedure:

1. Hypothesis Formulation: State the null hypothesis ($H_0$), which assumes no difference (e.g., $\mu = \mu_0$), and the alternative hypothesis ($H_a$).

   • Two-tailed: $H_a$ states the mean is different from $\mu_0$.
   • One-tailed: $H_a$ states the mean is specifically greater than or less than $\mu_0$. Select your significance level ($\alpha$), commonly set at 0.05.

2. Standard Error Calculation: Compute $SE = \frac{\sigma}{\sqrt{n}}$.

3. Z-score Calculation: Calculate the test statistic ($z$) to see how many standard errors the sample mean is from the hypothesized mean: $z = \frac{\bar{X} - \mu_0}{SE}$

4. Probability Determination: Find the p-value (the probability of observing a result as extreme as your test statistic) using a z-table or software.

5. Decision and Interpretation:

   • If $p \leq \alpha$, reject $H_0$. The result is statistically significant.
   • If $p > \alpha$, do not reject $H_0$.

Worked Example

Suppose a reference population of SAT scores has a mean ($\mu_0$) of 511 and a standard deviation ($\sigma$) of 120. A researcher tests a sample of 50 students ($n=50$) who had a mean score ($\bar{X}$) of 535. Using $\alpha = 0.05$:

• Step 1: $H_0: \mu = 511$; $H_a: \mu \neq 511$ (Two-tailed).
• Step 2: $SE = \frac{120}{\sqrt{50}} \approx 16.97$.
• Step 3: $z = \frac{535 - 511}{16.97} \approx 1.41$.
• Step 4: For $z = 1.41$ in a two-tailed test, the $p$-value is approximately 0.16.
• Step 5: Since $0.16 > 0.05$, we do not reject $H_0$. There is not enough evidence to say the course changed the scores.

Critical Caution: Statistical significance does not establish a causal relationship. Even if you reject the null hypothesis, the difference might be due to other factors in the research design, and it does not inherently prove "practical importance".