Non-Resistant Summary Statistics, Percentiles, Quartiles, and Boxplots

Limitations of Mean and Standard Deviation

• Single-number summary statistics provide critical insights into dataset characteristics, specifically central tendency (mean) and variability/dispersion (standard deviation).
• The mean and standard deviation have a major statistical limitation: they are not resistant to extreme values, outliers, or skewed tails.
• Extreme values heavily influence both the mean and standard deviation, pulling them toward the tail and failing to accurately represent the typical center or spread of the data.
• Non-resistant summary measures perform poorly when applied to non-symmetric distributions, such as:
  • Skewed distributions (histograms possessing a distinct tail to the left or right).
  • Bimodal or multimodal distributions (distributions that are not bell-shaped).
• When working with skewed datasets—such as real estate housing prices—position-based summary measures like the median and interquartile range (IQR) provide a far more accurate and stable representation of center and spread.
• Mastering resistant summary statistics completes the foundation for non-symmetric data analysis required prior to Exam 1.

The Median and Generalization to Percentiles

• Definition of the Median:
  • The median ($Q_2$) is a resistant measure of central tendency representing the exact positional center of an ordered dataset.
  • When observations are arranged in ascending order, the median divides the data into two equal halves:
  • Exactly $50\%$ of the data values fall at or below the median.
  • Exactly $50\%$ of the data values fall at or above the median.
• Percentile Generalization:
  • A percentile generalizes the concept of positional center to any arbitrary percentage $p$.
  • The $p\text{-th}$ percentile is defined as the numerical threshold where approximately $p\%$ of the dataset falls at or below that value, and the remaining $(100 - p)\%$ falls at or above it.
• Standardized Testing Case Study (SAT Performance):
  • Consider a student, Antonia, who scored $600$ on her SAT exam, placing her at the $74\text{th}$ percentile.
  • Interpretation: Exactly $74\%$ of all students who took the exam achieved a score less than or equal to her score of $600$
  • The remaining $26\%$ of test-takers scored equal to or greater than her score of $600$

Quartiles and the Five-Number Summary

• Quartiles:
  • Quartiles are three specific percentiles that divide an ordered dataset into four equal quarters, with each quarter containing $25\%$ of the observations.
  • First Quartile ($Q_1$): Equivalent to the $25\text{th}$ percentile. At $Q_1$, $25\%$ of the dataset lies below the value and $75\%$ lies above.
  • Second Quartile ($Q_2$): Equivalent to the $50\text{th}$ percentile, which is identically the Median. At $Q_2$, $50\%$ of the data lies below and $50\%$ lies above.
  • Third Quartile ($Q_3$): Equivalent to the $75\text{th}$ percentile. At $Q_3$, $75\%$ of the dataset lies below the value and $25\%$ lies above.
• The Five-Number Summary:
  • The Five-Number Summary provides an efficient, comprehensive snapshot of the shape, center, and spread of large datasets (ranging from thousands to tens of thousands of data points).
  • The Five-Number Summary consists of five key boundary values listed in ascending order:
  1. Minimum Value ($\text{Min}$): The smallest numerical observation in the dataset.
  2. First Quartile ($Q_1$): The $25\text{th}$ percentile threshold.
  3. Median ($Q_2$): The $50\text{th}$ percentile threshold / positional center.
  4. Third Quartile ($Q_3$): The $75\text{th}$ percentile threshold.
  5. Maximum Value ($\text{Max}$): The largest numerical observation in the dataset.

Manual and Algorithmic Calculation of Quartiles

• Manual Calculation Procedure:
  1. Arrange all raw numerical data points in ascending order.
  2. Locate the overall median ($Q_2$) of the ordered list.
  3. Divide the dataset into lower and upper halves using $Q_2$ as the dividing point.
  4. Calculate the median of the lower half to identify the first quartile ($Q_1$).
  5. Calculate the median of the upper half to identify the third quartile ($Q_3$).
• Sample Dataset Case Study (Ice Cream Bar Costs):
  • In a report evaluating ice cream bar market pricing, cost values were compiled for a sample size of $n = 27$ listed ice cream bars.
  • In a simplified data illustration with sample size $n = 13$, the minimum value observed was $0$
• Software and Calculator Discrepancies:
  • Different statistical computing software packages and graphing calculators may produce slightly different numerical values for $Q_1$ and $Q_3$
  • These minor variations occur because software programs utilize different internal algorithms and interpolation methodologies for discrete data point splits.
  • Variations resulting from standard hand calculations or standard calculator settings are acceptable and graded as correct on examinations.
  • For online practice assignments, if a calculated quartile is marked incorrect, utilize the platform's built-in graphing tool or exact algorithm options to align with its calculated thresholds.

Interquartile Range (IQR)

• Definition and Formula:
  • The Interquartile Range ($IQR$) is a resistant measure of spread that replaces standard deviation when analyzing skewed or non-bell-shaped distributions.
  • It measures the distance covered by the central portion of the dataset.
  • Formula for calculating the $IQR$:     $IQR = Q_3 - Q_1$
• Statistical Interpretation:
  • The $IQR$ represents the numerical span or range of the middle $50\%$ of the ordered dataset.
  • The lower $25\%$ of data falls below $Q_1$, the middle $50\%$ resides within the length of the $IQR$, and the upper $25\%$ lies above $Q_3$
• Understanding Concentration vs. Dispersion:
  • A small $IQR$ indicates that the middle $50\%$ of the data points are tightly concentrated around the median, indicating low variability.
  • A large $IQR$ indicates that even the central $50\%$ of observations is widely scattered, indicating high variability.

Construction and Visual Interpretation of Boxplots

• Boxplot Overview:
  • A boxplot is a visual representation of a two-dimensional distribution constructed directly from the Five-Number Summary.
• Step-by-Step Construction Method:
  1. Draw a continuous horizontal or vertical axis line with an appropriate scale encompassing all values from the minimum to maximum observations.
  2. Plot vertical line segments corresponding to the values of $Q_1$ and $Q_3$
  3. Construct a rectangular box bridging the line segment at $Q_1$ to the line segment at $Q_3$
  4. Draw a distinct line segment inside the box at the exact value of the median ($Q_2$).
  5. Extend line segments ("whiskers") outward from the edges of the box ($Q_1$ and $Q_3$) to endpoints marking the minimum ($\text{Min}$) and maximum ($\text{Max}$) values (e.g., extending from $111$ up to the maximum threshold).
• Visual Analysis of Spread:
  • The physical length of the central box visually equals the Interquartile Range ($IQR = Q_3 - Q_1$).
  • Boxplots provide an intuitive visual overview of dataset skewness, center, middle concentration, and extreme range limits.

Calculator Instructions and Homework Alignment

• Graphing Calculator Steps for Generating Boxplots:
  1. Open the statistical plotting interface by pressing Stat Plot.
  2. Select Plot 1 and switch the toggle setting to On.
  3. Select the boxplot icon from the list of available chart options.
  4. Adjust the viewing screen automatically by navigating to the Zoom menu and selecting ZoomStat to view the rendered plot.
• Practice Problem Alignment:
  • Practice problem sets directly match the structural design and question types found on homework assignments and examinations.
  • Individual practice exercises (such as Question 9 and Question 10) mirror homework questions with minor modifications to specific contextual details and numerical values.