Unit 4: Hypothesis Testing and Statistical Inferences

1. Concepts of Hypothesis

1.1 Definition and Nature

A Hypothesis is a tentative, testable statement about the relationship between two or more variables. In research, it serves as a predictive answer to a research question, subject to verification through statistical analysis.

1.2 Types of Hypotheses

• Null Hypothesis ():
  • The default position that there is no relationship between variables or no difference between groups.
  • Statistical tests are designed to reject or fail to reject .
  • Example: "There is no difference in test scores between Class A and Class B." ()
• Alternative Hypothesis ( or ):
  • The statement accepted if the null hypothesis is rejected. It suggests a significant relationship or difference exists.
  • Example: "Class A scores are different from Class B." ()

1.3 Directionality

• One-tailed Test (Directional): Predicts the specific direction of the effect. (e.g., "Class A performs better than Class B").
• Two-tailed Test (Non-directional): Predicts a difference but not the direction. (e.g., "Class A performs differently than Class B").

1.4 Errors in Hypothesis Testing

When making a decision about the Null Hypothesis, two types of errors can occur:

• Level of Significance (): The probability of making a Type I error (rejecting a true null). Standard levels are 0.05 (5%) and 0.01 (1%).
• P-value: The probability of obtaining results at least as extreme as the observed results, assuming is true.
  • If : Reject (Statistically Significant).
  • If : Fail to reject (Not Significant).

2. Parametric Tests

Definition: Statistical tests that make assumptions about the parameters of the population distribution from which the sample is drawn.

Key Assumptions:

1. Normality: Data follows a normal distribution (Bell curve).
2. Homogeneity of Variance: Variances between groups are approximately equal.
3. Scale: Dependent variable is Interval or Ratio scale.
4. Independence: Observations are independent of each other.

2.1 The Z-Test

Used to determine if there is a significant difference between sample and population means or between two sample means.

• Conditions for use:

  1. Sample size is large ().
  2. Population standard deviation () is known.
  3. Data is normally distributed.

• One-Sample Z-Test Formula:
  

  • Where: = Sample mean, = Population mean, = Population SD, = Sample size.

2.2 Student’s t-Test

Developed by William Sealy Gosset, this test compares means when sample sizes are small.

• Conditions for use:

  1. Sample size is small ().
  2. Population standard deviation () is unknown (sample standard deviation is used).
  3. Follows the t-distribution (flatter tails than normal distribution).

• Variations of the t-Test:

  1. One-Sample t-test: Compares a sample mean against a known population mean.
  2. Independent Samples t-test: Compares the means of two distinct, unrelated groups (e.g., Men vs. Women).
     • Formula:
       
  3. Paired Sample t-test (Dependent): Compares means from the same group at different times (e.g., Pre-test vs. Post-test).

2.3 The F-Test (ANOVA Basis)

The F-test is used to compare variances (Analysis of Variance - ANOVA). While t-tests compare means of two groups, F-tests/ANOVA are used for three or more groups.

• Concept: It compares the ratio of two variances.
  

• Interpretation:

  • If the variance between the groups is significantly larger than the variance within the groups, the F-ratio will be high, implying the means of the groups are different.
  • If (critical value), reject the Null Hypothesis.

3. Non-Parametric Tests

Definition: Statistical tests that do not assume a specific distribution (distribution-free) for the data. They are used when parametric assumptions are violated.

Key Characteristics:

• Used for Nominal (categorical) or Ordinal (ranked) data.
• Used when sample sizes are very small.
• Used when data is highly skewed (non-normal).

3.1 Chi-Square Test ()

Used to analyze categorical data to determine if observed frequencies differ from expected frequencies.

A. Chi-Square Goodness of Fit

• Purpose: Tests whether a sample distribution fits a theoretical population distribution.
• Example: Is a die fair? (Do numbers 1-6 appear with equal frequency?)

B. Chi-Square Test of Independence

• Purpose: Tests if two categorical variables are related or independent.

• Example: Is there an association between Gender (Male/Female) and Voting Preference (Republican/Democrat)?

• Formula:
  

  • = Observed frequency
  • = Expected frequency

• Degrees of Freedom (for Independence): .

3.2 Kruskal-Wallis Test (H-Test)

The non-parametric alternative to the One-Way ANOVA.

• Purpose: Used to compare differences between three or more independent groups when the dependent variable is ordinal or continuous but skewed (not normal).
• Procedure:
  1. Rank all data points from all groups together (1 = lowest score).
  2. Sum the ranks for each specific group.
  3. Calculate the H-statistic based on rank sums.
• Interpretation:
  • Null Hypothesis (): The population medians of all groups are equal.
  • If H is significant, at least one group dominates the others in terms of rank.

4. Summary Comparison: Parametric vs. Non-Parametric

5. Procedure for Hypothesis Testing (General Workflow)

1. Formulate Hypotheses: State and .
2. Select Level of Significance: Usually .
3. Choose the Test: Based on data type, number of groups, and distribution (e.g., t-test vs. Chi-square).
4. Calculate the Test Statistic: Compute the specific value (t-value, z-value, etc.).
5. Determine Critical Value/P-value: Consult statistical tables or software output.
6. Make a Decision: Compare calculated statistic to critical value (or p-value to alpha).
7. Conclusion: Interpret the result in the context of the research problem.