Statistical Tests - T-Test and Z-Test Notes
Chapter 9 Notes: Statistical Tests Overview
Single Sample Z-Test
- Formula:
z = \frac{M - \mu}{\sigma_M}
- Where:
- M = sample mean
- \mu = population mean
- \sigma_M is the standard deviation of the sample mean
Single Sample T-Test
- Purpose: Used when the population standard deviation (σ) is unknown.
- Estimated Standard Error:
- Formula:
s_M = \frac{s}{\sqrt{n}} - Where:
- s = sample standard deviation
- n = sample size
- Formula for T-Test:
t = \frac{M - \mu}{s_M}
Steps of the T-Test
- Set the hypotheses (null H0 and alternative H1 ).
- Set the decision criteria (select alpha level, directionality, critical t values).
- Compute the test statistic ( t ).
- Locate the test statistic in the sampling distribution and compare to critical value(s).
- Make a statistical decision (reject or fail to reject H_0 ).
- State the conclusion and describe the nature of the effect, if present.
Degrees of Freedom
- For a single sample t-test:
df = n - 1
Finding Critical T-Values
- Use Table B2 for the t-distribution.
- Example:
- Sample size: n = 4
- One-tailed test; \alpha = .05
- df = 3 ; critical value t = 2.353
- Two-tailed test example:
- Sample size: n = 9 ; df = 8 ;
- Critical values t = \pm 2.306
Influence of Sample Variance and Size on T-Test Outcome
- Larger sample variance → Larger standard error → Smaller t → Lower power
- Smaller sample variance → Smaller standard error → Larger t → Higher power
- Smaller sample size → Larger standard error → Smaller t → Lower power
- Larger sample size → Smaller standard error → Larger t → Higher power
- Power: Probability of correctly rejecting the null hypothesis.
Assumptions of the Single Sample T-Test
- Independent observations.
- The population sampled must be normally distributed.
When to Use Z-Test vs. T-Test
- Check for variability information for the population:
- If information is present (e.g., SS , \sigma^2 , or \sigma ), use a z-test.
- If not, use a t-test.
Chapter 9 Examples
Example 1: Police Officer Work Hours
- Claim: Officers work significantly more than 40 hours.
- Data: 9 officers, hours worked: 43, 45, 48, 41, 45, 50, 50, 53, 48
- Calculations:
- n=9 , s^2=14.5 , M=47 , s=3.81
- Hypotheses:
- H_0: \mu = 40
- H_1: \mu > 40
- \alpha = .05 , one-tailed
- df = 8
- Critical t = +1.860
- Test statistic:
t = \frac{47 - 40}{1.27} = 5.51
- Result: Reject H_0 ; officers work significantly more than 40 hours, t(8) = 5.51, p < .05 .
Example 2: Study Skills Training Program
- Claim: Students in the program perform differently than the rest of the class.
- Data: 25 freshmen, past class mean = 74, program mean = 78, SS=2400 .
- Calculations:
- n=25 , s^2=100 , M=78 , s=10
- Hypotheses:
- H_0: \mu = 74
- H_1: \mu > 74
- \alpha = .05 , one-tailed
- df = 24
- Critical t = +1.711
- Test statistic:
t = \frac{78 - 74}{2} = 2
- Result: Reject H_0 ; students in the program performed better, t(24) = 2.0, p < .05 .
Example 3: Humidity Effect on Rats' Eating
- Claim: Humidity affects eating behavior.
- Data: $ n=100 $, sample mean = 18.7, population mean = 21, SS=2475 .
- Calculations:
- Hypotheses:
- H_0: \mu = 21
- H_1: \mu \neq 21
- \alpha = .05 , two-tailed
- df = 99 , critical t = \pm 2.00
- Test statistic:
t = \frac{18.7 - 21}{0.5} = -4.6
- Result: Reject H_0 ; humidity significantly decreases food consumption, t(99) = -4.6, p < .05 .