Contingency Tables, Marginal Distributions, and Conditional Distributions in Statistics

Contingency Tables, Marginal Distributions, and Conditional Distributions

Contingency Tables

Contingency tables are a fundamental tool in statistics for summarizing the relationship between two categorical variables. They display the frequency (or count) of observations that fall into each combination of categories.

• Definition: A contingency table (also called a cross-tabulation or crosstab) is a matrix that shows the frequency distribution of variables.

• Structure: Rows typically represent one categorical variable, while columns represent another.

• Purpose: Used to analyze the association between variables and to compute marginal and conditional distributions.

• Example: A table showing the number of drivers who use cell phones and have speeding violations.

Marginal Distribution

Marginal distributions describe the totals for each category of a single variable, ignoring the other variable. They are obtained by summing across rows or columns of a contingency table.

• Definition: The marginal distribution of a variable is the distribution of totals for that variable, found by summing across the other variable.

• Calculation: Add the frequencies in each row or column to get the marginal totals.

• Relative Marginal Distribution: Divide each marginal total by the overall total to get relative frequencies.

• Formula: $\text{Relative Frequency} = \frac{\text{Row or Column Total}}{\text{Grand Total}}$

• Example: In a table of drivers, the marginal distribution for 'Uses cell phone while driving' is the total number of such drivers divided by the overall total.

Conditional Distribution

Conditional distributions show the distribution of one variable for a fixed value of the other variable. They are useful for understanding how the distribution of one variable changes depending on the category of another.

• Definition: The conditional distribution of a variable is the distribution of that variable for a specific value of the other variable.

• Calculation: Divide each cell frequency by the total for the relevant row or column (depending on which variable is conditioned upon).

• Formula: $\text{Conditional Frequency} = \frac{\text{Cell Frequency}}{\text{Row or Column Total}}$

• Example: The proportion of drivers with speeding violations among those who use cell phones while driving.

Worked Examples and Tables

Example 1: Cell Phone Use and Speeding Violations

	Speeding Violation in Last Year	No Speeding Violation in Last Year	Frequency (Total)
Uses cell phone while driving	25	280	305
Does not use cell phone while driving	45	405	450
Frequency (Total)	70	685	755

Marginal Distribution (Relative Frequency)

• Uses cell phone while driving: $\frac{305}{755} = 0.4040$

• Does not use cell phone while driving: $\frac{450}{755} = 0.5960$

• Total: $1.0000$

Conditional Distribution (Speeding Violation)

• Among those who use cell phones while driving: $\frac{25}{305} = 0.0820$ (Speeding Violation)

• Among those who do not use cell phones while driving: $\frac{45}{450} = 0.1000$ (Speeding Violation)

• Additional info: Conditional distributions can be calculated for either rows or columns, depending on the context.

Example 2: Handedness and Gender

	Right-handed	Left-handed	Frequency (Total)
Males	66	15	81
Females	45	9	54
Frequency (Total)	111	24	135

Marginal Distribution (Relative Frequency)

• Right-handed: $\frac{111}{135} = 0.8222$

• Left-handed: $\frac{24}{135} = 0.1778$

• Total: $1.0000$

Conditional Distribution (Gender)

• Among males: $\frac{66}{81} = 0.8148$ (Right-handed), $\frac{15}{81} = 0.1852$ (Left-handed)

• Among females: $\frac{45}{54} = 0.8333$ (Right-handed), $\frac{9}{54} = 0.1667$ (Left-handed)

Example 3: Opinions by Ethnicity

	Good thing	Bad thing	Good and Bad	No Opinion	Relative Frequency
White Respondents	160	110	9	9	$\frac{288}{785} = 0.3674$
Black Respondents	152	35	12	9	$\frac{208}{785} = 0.2650$
Hispanic Respondents	144	33	12	10	$\frac{199}{785} = 0.2532$
Total	456	178	33	28	$1.0000$

Marginal Distribution (Relative Frequency)

• Good thing: $\frac{456}{785} = 0.5815$

• Bad thing: $\frac{178}{785} = 0.2268$

• Good and Bad: $\frac{33}{785} = 0.0420$

• No Opinion: $\frac{28}{785} = 0.0357$

• Total: $1.0000$

Conditional Distribution (by Ethnicity)

• For each ethnic group, divide each cell by the row total to get the conditional distribution.

• Example for White Respondents: $\frac{160}{288} = 0.5556$ (Good thing), $\frac{110}{288} = 0.3819$ (Bad thing), etc.

Summary Table: Types of Distributions in Contingency Tables

Type	Definition	Calculation	Purpose
Marginal Distribution	Distribution of totals for one variable	Sum across rows or columns	Describes overall distribution of a variable
Conditional Distribution	Distribution of one variable for a fixed value of another	Divide cell by row or column total	Shows how distribution changes with another variable

Key Points

• Contingency tables organize data for two categorical variables.

• Marginal distributions summarize totals for each variable.

• Conditional distributions show the distribution of one variable given a specific value of another.

• Relative frequencies are calculated by dividing counts by the grand total.

• Conditional frequencies are calculated by dividing counts by the relevant row or column total.

Additional info: These concepts are foundational for understanding statistical association, independence, and for performing chi-square tests of independence.