Hypothesis Testing: Concepts, Steps, and Applications

Hypothesis Testing

Introduction to Inferential Statistics

Inferential statistics involves making conclusions about a population based on sample data. Two main approaches are estimation (such as confidence intervals) and hypothesis testing.

• Estimation: Provides interval estimates for population parameters (e.g., mean, proportion).

• Hypothesis Testing: Assesses claims about population parameters using sample evidence and probability theory.

Estimation and Confidence Intervals

When the population mean ($\mu$) is unknown, we estimate it using a sample mean ($\bar{X}$) and construct a confidence interval:

• Confidence Interval Formula:

$\bar{X} \pm Z \frac{\sigma}{\sqrt{n}}$

• Example: "We are 95% confident that the population mean lies between 140kg and 150kg."

What is a Hypothesis?

Definition and Examples

A hypothesis is a statement about the value of a population parameter, developed for testing. It is specific and testable.

• Examples:

• The mean monthly income for system analysts is $3,625.

• 90% of all income tax statements are filled correctly.

• The mean weight of tyres produced by Dunlop is 56kg.

What is Hypothesis Testing?

Hypothesis testing is a procedure based on sample evidence and probability theory to determine whether a hypothesis about a population parameter should be rejected or not.

• Uses sample data to make probabilistic decisions about population parameters.

• Forms the core of inferential statistics.

Steps in Hypothesis Testing

Overview of the Five Steps

1. State the null and alternative hypotheses.

2. Select a level of significance ($\alpha$).

3. Identify the test statistic.

4. Formulate the decision rule.

5. Take a sample and arrive at a decision.

Step 1: State the Null and Alternative Hypotheses

• Null Hypothesis ($H_0$): The default assumption; usually states "no effect" or "no difference." Always contains an equality (e.g., $\mu = 70$).

• Alternative Hypothesis ($H_1$ or $H_a$): The research hypothesis; what the researcher aims to support. It is the opposite of $H_0$ and never contains an equality (e.g., $\mu \neq 70$).

Examples:

• $H_0$: The mean strength of steel is NOT significantly different from 70 psf ($\mu = 70$).

• $H_1$: The mean strength of steel is significantly different from 70 psf ($\mu \neq 70$).

• $H_0$: There is NO difference in mean usable life between Eveready and X-brand batteries ($\mu_1 = \mu_2$).

• $H_1$: There is a difference in mean usable life between Eveready and X-brand batteries ($\mu_1 \neq \mu_2$).

Directional vs. Non-directional Hypotheses

• Non-directional (Two-tailed): $H_1: \mu \neq \mu_0$

• Directional (One-tailed): $H_1: \mu > \mu_0$ or $H_1: \mu < \mu_0$

Step 2: Select a Level of Significance ($\alpha$)

The level of significance ($\alpha$) is the probability of making a Type I error (rejecting $H_0$ when it is true). Common values are 0.05, 0.01, or 0.10.

• Example: $\alpha = 0.05$ means a 5% risk of Type I error.

• Lower $\alpha$ means stricter criteria for rejecting $H_0$.

Step 3: Identify the Test Statistic

• The choice of test statistic depends on the objective and data type (e.g., Z-test, t-test, ANOVA, Chi-Square, Regression).

• The value of the test statistic is used to decide whether to reject $H_0$.

Step 4: Formulate the Decision Rule

• Determine the critical value(s) based on $\alpha$ and the type of test (one-tailed or two-tailed).

• One-tailed test: All of $\alpha$ in one tail (e.g., $Z_c = 1.645$ for $\alpha = 0.05$).

• Two-tailed test: $\alpha$ split between both tails (e.g., $Z_c = 1.96$ for $\alpha = 0.05$).

• If the test statistic falls in the rejection region, reject $H_0$; otherwise, fail to reject $H_0$.

Step 5: Take a Sample and Arrive at a Decision

• Collect sample data, compute the test statistic, and compare it to the critical value.

• Make a decision to reject or fail to reject $H_0$ based on the comparison.

Types of Errors in Hypothesis Testing

• Type I Error (False Positive): Rejecting $H_0$ when it is true. Probability is $\alpha$.

• Type II Error (False Negative): Failing to reject $H_0$ when it is false. Probability is $\beta$.

Consequences:

• Type I Error: May lead to unnecessary actions (e.g., unnecessary medical treatment).

• Type II Error: May miss a real effect (e.g., failing to diagnose a disease).

Decision	Reality: $H_0$ True	Reality: $H_0$ False
Accept $H_0$	Correct Decision	Type II Error
Reject $H_0$	Type I Error	Correct Decision

Examples of Hypothesis Testing

Testing a Population Mean (Known $\sigma$)

• Example: Test if the mean efficiency rating of Boeing employees is still 200 ($\sigma = 16$, $n = 100$, $\alpha = 0.01$).

• Test Statistic:

$Z = \frac{\bar{X} - \mu}{\sigma / \sqrt{n}}$

• Compare $Z$ to critical value ($Z_c = \pm 2.58$ for $\alpha = 0.01$ two-tailed).

Testing a Population Mean (Unknown $\sigma$)

• Use sample standard deviation $s$ if $n > 30$.

• Example: Credit manager tests if mean unpaid balance is more than $400 ($n = 172$,$\bar{X} = 407$,$s = 38$,$\alpha = 0.05$).

• Test Statistic:

$Z = \frac{\bar{X} - \mu}{s / \sqrt{n}}$

• Compare $Z$ to $Z_c = 1.645$ (one-tailed).

Testing the Difference Between Two Means

• For independent samples:

$Z = \frac{\bar{X}_1 - \bar{X}_2}{\sqrt{\frac{s_1^2}{n_1} + \frac{s_2^2}{n_2}}}$

• Example: Test if there is a difference in average heights of adult females from two countries.

Summary Table: Types of Hypothesis Tests

Test	Parameter	Null Hypothesis	Alternative Hypothesis	Test Statistic
One-sample Z-test	Mean ($\mu$)	$\mu = \mu_0$	$\mu \neq \mu_0$	$Z = \frac{\bar{X} - \mu_0}{\sigma / \sqrt{n}}$
Two-sample Z-test	Means ($\mu_1, \mu_2$)	$\mu_1 = \mu_2$	$\mu_1 \neq \mu_2$	$Z = \frac{\bar{X}_1 - \bar{X}_2}{\sqrt{\frac{s_1^2}{n_1} + \frac{s_2^2}{n_2}}}$

Key Points to Remember

• Always state hypotheses in terms of population parameters.

• The null hypothesis always contains the equality sign.

• Type I and Type II errors are inherent risks in hypothesis testing.

• The choice of test and significance level depends on the research question and data type.

Additional info: This guide covers the fundamental steps and concepts of hypothesis testing, including error types, test statistics, and decision rules, suitable for undergraduate statistics students.