SDS CH3 - Descriptive Statistics: Five-Number Summaries, Box Plots, and the Empirical Rule

The five-number summary is a foundational descriptive tool used to quickly assess the distribution of a dataset. It consists of five specific values: - 1. Minimum - 2. First Quartile ( $Q_1$ ) - 3. Second Quartile ( $Q_2$ ), also known as the Median - 4. Third Quartile ( $Q_3$ ) - 5. Maximum
Relationships within the five-number summary indicate the skewness of the data. Specifically, the differences between these values determine the skew: - Left-Skewed Data (Negative Skew): - The distance between the median and the minimum is greater than the distance between the maximum and the median ( $\text{Median} - \text{Min} > \text{Max} - \text{Median}$ ). - The distance between $Q_1$ and the minimum is greater than the distance between the maximum and $Q_3$ ( $Q_1 - \text{Min} > \text{Max} - Q_3$ ). - The longer tail is on the left side of the distribution, which pulls the mean in that direction. - Symmetric Data: - The distances between the median and the extremes are approximately equal, reflecting a balanced distribution centered around the median. - Right-Skewed Data (Positive Skew):
A critical resource for visualizing these distances is the Slide 57 diagram found under Chapter 3 of the SunLearn page. This diagram serves as a companion to the summary table by illustrating how these mathematical distances manifest visually in a distribution.

A box plot (also known as a box-and-whisker plot) displays the five-number summary on an ordered number line.
Structural Components: - The Box: Represents the middle $50\%$ of the data, extending from $Q_1$ to $Q_3$ . This range is the Interquartile Range ( $IQR$ ). - The Median Line: A vertical line inside the box that splits the middle $50\%$ into two segments of $25\%$ each. - The Whiskers: Lines extending from the box out to the minimum and maximum values, representing the lower and upper $25\%$ of the data, respectively.
Interpreting Skewness in Box Plots: - In a Left-Skewed plot, the left whisker is significantly longer than the right whisker. - In a Symmetrical plot, the median is centered in the box, and both whiskers are of equal length. - In a Right-Skewed plot, the right whisker is longer, pulling the mean to the right (where $\text{Mean} > \text{Median}$ ).
Orientation and Comparison: - Box plots can be oriented horizontally or vertically. - Vertical orientations are frequently used to facilitate comparison between multiple groups (e.g., comparing "Group 1" and "Group 2" side-by-side) to observe differences in medians, variation, and overall distribution patterns.

While digital tools like Excel are typically used for construction, understanding the manual process provides intuition. Consider the following dataset values: - $\text{Minimum} = 0$ - $Q_1 = 2$ - $Q_2 = 3$ - $Q_3 = 5$ - $\text{Maximum} = 27$
Construction Steps: - 1. Draw a number line encompassing the range (e.g., $0$ to $30$ with intervals of $5$ ). - 2. Mark the minimum at $0$ and the maximum at $27$ . - 3. Draw the box limits at $Q_1 = 2$ and $Q_3 = 5$ . - 4. Draw the median line inside the box at $Q_2 = 3$ .
Observation: This specific distribution is right-skewed (positively skewed) because the right tail (from $5$ to $27$ ) is significantly longer than the left tail (from $0$ to $2$ ).

The study of descriptive statistics is categorized into three primary areas: - 1. Central Tendency (Average Inclination): - Mean ( $\bar{x}$ or $\mu$ ) - Median - Mode - Geometric Mean and Geometric Mean Rates of Return - 2. Spread or Variation (Deviation from Center): - Range - Variance - Standard Deviation - Coefficient of Variation - Interquartile Range ( $IQR$ ): Considered a "resistant measure" of variation because it is not influenced by extreme outliers. - 3. Shape (Patterns of Distribution): - Skewness (Direction of the tail) - Kurtosis (Peakedness of the distribution) - Five-Number Summary and Box Plots
Note on Notation: Sigma notation ( $\sum$ ) is not a separate topic but a universal tool used throughout all these calculations.

The "Big Circle, Small Circle" metaphor illustrates the relationship between a population and a sample: - Population (Big Circle): The entire group containing the characteristic of interest. It is often too large or resource-intensive to study in its entirety. - Sample (Small Circle): A manageable subset of the population. It must be representative of the population for findings to be valid.
Parameters vs. Statistics: - Parameters: Summary measures derived from the population. These represent the "ground truth" and are unchanging, yet usually unknown. - Statistics: Summary measures derived from the sample. These are estimators of the parameters, representing a "best guess" available based on limited data. - Statistics vary depending on the sample collected due to the nature of random sampling, leading to inherent uncertainty.

Formulas differ slightly depending on whether one is calculating a population parameter or a sample statistic.
Notation Rules: Sample measures use Latin letters; Population measures use Greek letters.

Sample Mean ( $\bar{x}$ ): $\bar{x} = \frac{1}{n} \sum_{i=1}^{n} x_i$
Population Mean ( $\mu$ ): $\mu = \frac{1}{N} \sum_{i=1}^{N} x_i$ Note: $N$ represents the total population size; $n$ represents sample size.

Sample Variance ( $s^2$ ): $s^2 = \frac{1}{n-1} \sum_{i=1}^{n} (x_i - \bar{x})^2$
Population Variance ( $\sigma^2$ ): $\sigma^2 = \frac{1}{N} \sum_{i=1}^{N} (x_i - \mu)^2$
The Bias Correction ( $n-1$ ): - The sample variance tends to underestimate the true population variance. - Dividing by $n-1$ instead of $n$ serves as a clever correction to slightly increase the result, fixing this underestimation. - The effect of this correction is more dramatic for small sample sizes; as $n$ increases, the adjustment relative to the size of $n$ becomes smaller.

Sample Standard Deviation ( $s$ ): $s = \sqrt{s^2} = \sqrt{\frac{1}{n-1} \sum_{i=1}^{n} (x_i - \bar{x})^2}$
Population Standard Deviation ( $\sigma$ ): $\sigma = \sqrt{\sigma^2} = \sqrt{\frac{1}{N} \sum_{i=1}^{N} (x_i - \mu)^2}$

The Empirical Rule is a "rule of thumb" used to describe the data distribution in symmetric, bell-shaped distributions.
It establishes the approximate percentage of data contained within certain standard deviations from the mean: - 68% of data falls within $\mu \pm 1\sigma$ . - 95% of data falls within $\mu \pm 2\sigma$ . - 99.7% of data falls within $\mu \pm 3\sigma$ .
Outliers: The remaining $0.3\%$ of the data (beyond three standard deviations) represents points with extremely low probability/likelihood, categorized as outliers.
Crucial Constraint: The Empirical Rule cannot be applied to skewed or non-bell-shaped distributions; it relies entirely on the principle of symmetry.

Question Regarding Terminology: A student asked about the term "negatively skewed" encountered in other materials.
Response: "Negatively skewed" is an interchangeable term for "left-skewed." Conversely, "positively skewed" is interchangeable with "right-skewed." This correlates with the direction on a standard number line where negative values are to the left and positive values are to the right.