AP Statistics Unit 8: Inference for Categorical Data

Chi-Square ($\chi^2$) procedures

Designed for non-binary categorical data involving counts.

Null Hypothesis ($H_0$)

The hypothesis that the categorical variable has a specified distribution.

Chi-Square statistic formula

\chi^2 = \sum \frac{(O - E)^2}{E}

Observed count ($O$)

The actual count observed in the sample data.

Expected count ($E$)

The count predicted by the null hypothesis.

Degrees of Freedom ($df$)

A parameter related to the number of categories or groups, affecting the Chi-Square distribution.

Right-tailed test

Chi-Square tests are inherently right-tailed, focusing on values greater than expected.

Goodness-of-Fit test

Used when testing if a sample distribution differs from a proposed distribution.

$p_i$ in chi-square

The hypothesized proportion for a category.

Expected Counts Condition

All expected counts must be at least 5 for a valid Chi-Square test.

Chi-Square test for Homogeneity

Determines if the distribution of a categorical variable is the same across multiple populations.

Chi-Square test for Independence

Used to determine if there is an association between two categorical variables.

Mechanics for calculating expected counts

Ei = n \times pi where $n$ is total sample size.

Large Counts Condition

Condition stating that all expected counts must be at least 5.

Chi-Square statistic calculation example

Using the formula to compute $\chi^2$ based on observed and expected counts.

P-value in Chi-Square tests

The probability of obtaining a Chi-Square statistic as extreme as the observed value, assuming $H_0$ is true.

Homogeneity test hypotheses

$H0$: The true distribution is the same across all groups. $Ha$: The distribution is different for at least two groups.

Null vs Alternative Hypotheses

$H0$: No difference; $Ha$: There is a difference.

Chi-Square distribution characteristics

Non-negative, skewed right, defined by degrees of freedom.

Chi-Square goodness-of-fit test

Tests if a single categorical variable follows a specific distribution.

Degrees of freedom formula for contingency tables

df = (r-1)(c-1) where $r$ is number of rows and $c$ is number of columns.

Chi-Square statistic interpretation

Value indicating the discrepancy between observed and expected counts.

Common mistake using counts

Using proportions instead of raw counts in Chi-Square calculations.

Expected counts must be ≥ 5

A requirement for valid Chi-Square goodness-of-fit and homogeneity tests.

Causation and Chi-Square association

Association does not imply causation; Chi-Square tests show relationship, not cause-effect.

Misinterpretation of alternative hypothesis

Stating all groups are different instead of at least two groups differing.

Check for random sampling condition

Ensure data comes from a random sample for Chi-Square tests.

Chi-Square test example

Illustrates testing liquor store distribution against theoretical proportions.

Comparative experiments in Homogeneity tests

Arise when assessing distributions across multiple treatment groups.

P-value interpretation in hypothesis testing

Used to decide whether to reject or fail to reject the null hypothesis.

Goodness-of-fit test purpose

To verify if sample distribution matches a theoretical model.

Expected counts formula for specific cells

E = \frac{\text{Row Total} \times \text{Column Total}}{\text{Table Total}}

Evidence for $H_0$ in tests

A small P-value suggests rejecting the null hypothesis.

Chi-Square distribution family properties

Flattens and becomes more symmetric with increasing degrees of freedom.

Significance level in hypothesis testing

Commonly set at $\alpha = 0.05$ for determining statistical significance.

Independence test structure

Examining relationships between two categorical variables within one population.

Relationship between variables

Chi-Square tests explore association, not causation, between categorical variables.

Calculating Chi-Square statistic

Involves sums of squared differences between observed and expected counts.

Sufficient evidence conclusion

Rejecting $H_0$ indicates strong evidence against the null hypothesis.

Conditions for Chi-Square tests

Check random sampling, expected counts, and proportion conditions.

Using Chi-Square distribution tables

Refer to tables to find critical values based on df and significance level.

Limitations of Chi-Square tests

Cannot be used for small sample sizes or if expected counts are too low.

Chi-Square goodness-of-fit graphical representation

Visual representation of expected vs. observed counts.

Sample sizes considered in tests

Assessments depend on adequate sample sizes for validity in tests.

Chi-Square application scenarios

Used in various fields, including health, marketing, and social sciences.

Chi-Square for contingency tables analysis

Analyzes relationships and associations across two categorical variables.