Stat 240/250 Midterm Study Guide: Describing and Analyzing Quantitative Data & Regression

Describing Data: The Five W's and Variable Types

Cases and Variables

Understanding the structure of a dataset begins with identifying the cases (the subjects or units of analysis) and the variables (the characteristics measured).

• Who: Refers to the cases or subjects being studied.

• What: Refers to the variables measured for each case.

Variable Types

• Quantitative Variables: Variables that represent numerical values and can be measured or counted (e.g., height, weight).

• Categorical Variables: Variables that represent categories or groups (e.g., gender, color).

Case: An individual unit or subject in the dataset.

Describing Quantitative Variables: Graphs and Numerical Summaries

Graphical Representations

• Histogram: A graphical display of the distribution of a quantitative variable. Shows shape, center, and spread.

Measures of Center

• Mean: The arithmetic average. Sensitive to outliers.

• Median: The middle value when data are ordered. Resistant to outliers.

Percentiles and Five-Number Summary

• Percentiles: Values below which a certain percent of data fall. The five-number summary uses specific percentiles.

• Five-Number Summary: Minimum, Q1 (25th percentile), Median (50th percentile), Q3 (75th percentile), Maximum.

Measures of Spread

• Standard Deviation (SD): Measures average distance from the mean. Nonresistant to outliers.

• Interquartile Range (IQR): Difference between Q3 and Q1. Resistant to outliers.

Choosing Statistics

• Use mean and SD for symmetric distributions without outliers.

• Use median and IQR for skewed distributions or with outliers.

Resistant vs. Nonresistant Statistics

• Resistant: Not affected by extreme values (e.g., median, IQR).

• Nonresistant: Sensitive to extreme values (e.g., mean, SD).

Identifying Shape

• Four ways: symmetry/skewness, modality (number of peaks), outliers, spread.

Boxplots and Outlier Detection

Boxplots

• Visualize the five-number summary.

• Can compare distributions side-by-side.

• Show shape, center, spread, and outliers.

Outlier Identification

• Outliers are values that fall outside typical boundaries.

• Common boundaries: values beyond ±2 SD (moderate outliers), beyond ±3 SD (serious outliers).

Explanatory vs. Response Variables

• Explanatory (Predictor): Variable used to explain or predict another.

• Response: Variable being predicted or explained.

Z-Scores

• Standardizes values for comparison.

• Formula: $z = \frac{x - \mu}{\sigma}$

• Interpretation: How many SDs a value is from the mean.

Empirical Rule

• Applies to normal distributions.

• About 68% of data within ±1 SD, 95% within ±2 SD, 99.7% within ±3 SD.

Describing Two Quantitative Variables: Scatterplots and Correlation

Scatterplots

• Graphical display of two quantitative variables.

• Shows pattern, direction, strength, and outliers.

Correlation

• Measures strength and direction of linear relationship.

• Range: -1 to 1.

• Formula: $r = \frac{\sum (x_i - \bar{x})(y_i - \bar{y})}{(n-1)s_x s_y}$

• Interpretation: Positive/negative, strong/weak.

Explanatory vs. Response Variables

• Explanatory (predictor) variable on x-axis; response variable on y-axis.

Simple Linear Regression Model

• Equation: $\hat{y} = b_0 + b_1 x$

• Slope (b1): Change in y per unit change in x.

• Intercept (b0): Predicted y when x = 0.

• Find predicted value $\hat{y}$ by substituting x.

Regression Model Variation and Diagnostics

Least Squares Method

• Estimates slope and intercept by minimizing sum of squared residuals.

• Residual: $e_i = y_i - \hat{y}_i$

Analysis of Variance (ANOVA) Table

• Summarizes sources of variation in regression.

• Components: Regression, Residual/Error, Total.

• Generic ANOVA table:

Source	DF	SS	MS	F
Regression	1	SSR	MSR	F
Residual	n-2	SSE	MSE
Total	n-1	SST

Quantity Minimized

• Least Squares minimizes $\sum (y_i - \hat{y}_i)^2$

R2 (Coefficient of Determination)

• Proportion of variance in y explained by x.

• Formula: $R^2 = 1 - \frac{SSE}{SST}$

• Interpretation: Higher R2 means better fit, but does not guarantee best model.

Standard Error of Residuals (se)

• Measures typical size of residuals.

• Compare se to sy (SD of y) to assess prediction quality.

Regression Model Conditions and Diagnostics

Model Conditions

• Check graphs for linearity, constant variance, independence, and normality of residuals.

• Meeting conditions is distinct from model's predictive ability.

Regression Diagnostics: Unusual Observations

• Y-axis (R flags): Outliers in response variable.

• X-axis (X flags): Outliers in predictor variable.

• Highly Influential Observations: Points that greatly affect regression estimates.

Outlier Boundaries and Labels

• Moderate outliers: Beyond ±2 SD.

• Serious outliers: Beyond ±3 SD.

Summary and Integration

Bringing Concepts Together

• Apply concepts of descriptive statistics and regression to analyze real datasets.

• Use worksheets and practice problems to reinforce understanding.

Example: Given a dataset of heights and weights, describe the distribution of heights using histograms, five-number summary, and boxplots. Then, analyze the relationship between height and weight using scatterplots, correlation, and linear regression.

Additional info: Academic context and formulas have been expanded for clarity and completeness.