Data Distributions and Graphical Summaries

Introduction to Distributions

Understanding the distribution of a dataset is a foundational concept in statistics. A distribution records all observed values from a sample and how frequently each value occurs. This process helps in visualizing and summarizing data to identify patterns, central tendencies, and variability.

• Distribution of a Sample: Describes the values a variable takes and their frequencies.

• Frequency: The count of how many times a value appears in the dataset.

• Two-Step Investigative Process:

  1. Visualize the distribution using an appropriate graphic.

  2. Summarize the distribution by describing its center, spread, and shape.

Dotplots

A dotplot is a simple graphical summary for numerical data, especially effective for small datasets. Each observation is represented by a dot placed above its value on a number line.

• Shows individual data points and patterns.

• Best for small datasets; can become cluttered with large datasets.

• Useful for detecting clusters, gaps, and outliers.

• Example: Number of days absent: 0, 0, 1, 2, 2, 2, 3, 3, 3, 4, 4, 5, 6 (see Example B below).

Histograms

A histogram is a graphical summary for numerical data that groups observations into continuous intervals called bins. It provides a smoothed view of the data distribution compared to dotplots.

• Number line divided into bins of equal width.

• Vertical bars represent the frequency (or relative frequency) of observations in each bin.

• Bin Width Effects:

  • Narrow bins: High detail, but may be too "spiky".

  • Wide bins: Smoother, but may hide important details.

  • Choosing appropriate bin width is essential.

• Boundary Rules: When a value falls on a bin boundary, a consistent rule (e.g., right-hand rule) is used to assign it to a bin.

Relative Frequency Histograms

A relative frequency histogram displays the proportion or percentage of observations in each bin, rather than raw counts. This changes the vertical scale but not the overall shape of the histogram.

• Relative frequency for a bin: (or 75%)

• Helps compare distributions across different sample sizes.

Visualizing Categorical Variables

Bar Charts (Bar Graphs/Bar Plots)

A bar chart is used to display the distribution of categorical variables. Each category is represented by a bar, with height proportional to its frequency or relative frequency.

• Bars are separated by gaps to indicate distinct categories.

• Categories can be ordered by frequency (Pareto chart) or in a meaningful sequence.

• Shape remains the same whether using counts or proportions; only the vertical axis changes.

Pareto Charts

A Pareto chart is a specialized bar chart where categories are ordered from most to least frequent. Named after economist Vilfredo Pareto, it is useful for highlighting dominant categories.

Pie Charts

A pie chart displays relative frequencies as slices of a circle. Each wedge's area is proportional to the category's relative frequency.

• Rarely used in scientific analysis due to difficulty in comparing wedge areas and clutter with many categories.

Bar Charts vs. Histograms

• Bar Widths: Arbitrary in bar charts; meaningful (bin width) in histograms.

• Gaps: Bar charts have gaps to show distinct categories; histograms have gaps only if no data in a bin.

Describing Distributions

Numerical Variables

• Shape: Symmetric, right-skewed, left-skewed, unimodal, bimodal, or multimodal.

• Center: Typical value (mean, median, or mode).

• Spread: Variability or dispersion of data.

• Outliers: Data points far from the rest, requiring further investigation.

• Multi-group Data: Bimodal or multimodal distributions may indicate multiple groups combined.

Categorical Variables

• Mode: Category with the highest frequency (can be bimodal or multimodal).

• Variability (Diversity): Degree to which observations are spread across categories.

  • High variability: Observations spread across many categories.

  • Low variability: Most observations in a single category.

Measures of Center and Spread

Overview

Measures of center and spread summarize the typical value and variability in a dataset. The most common measures are the mean, median, mode, and standard deviation.

• Mean: Arithmetic average of all values.

• Median: Middle value when data are ordered.

• Mode: Most frequently occurring value.

• Standard Deviation: Measures the typical distance of observations from the mean.

Standard Deviation Formula

The sample standard deviation () is calculated as:

• Find deviation:

• Square deviations:

• Sum squared deviations:

• Divide by degrees of freedom:

• Take square root to return to original units.

Weighted Mean and Frequency Distribution Mean

• Weighted Mean:

• Frequency Distribution Mean:

• Where is the value, is the weight, and is the frequency.

Calculator Instructions (TI-84)

• To find the mean and standard deviation:

  1. STAT → Edit → Enter data into L1.

  2. STAT → CALC → 1-Var Stats → ENTER.

  3. Mean is ; sample standard deviation is .

Problem-Solving Examples

Example A: Weighted Mean

• Given categories with weights and scores, the weighted mean is calculated as .

• Example: Weighted mean = 74.

Example B: Dot Plot Analysis (Number of Days Absent)

• Data: 0, 0, 1, 2, 2, 2, 3, 3, 3, 4, 4, 5, 6

• Mean: 2.69

• Median: 3

• Mode: 3

Example C: Alex Rodriguez (A-Rod) Home Run Dataset

• Raw data (sorted): 0, 5, 9, 9, 16, 18, 23, 36, 35, 30, 35, 54, 35, 36, 41, 42, 42, 48, 52, 57, 35, 36...

• Mean: 30.30

• Median: 35

• Mode: Multi-modal (36, 0, 42, 35, 30)

• Histogram shape: Bimodal (two peaks: mid-30s/40s and lower-frequency seasons).

Interpreting and Avoiding Misleading Graphs

Misleading Graphs

• Altering the vertical axis (not starting at zero) exaggerates differences between bars.

• Using area instead of height (e.g., pictograms) can mislead, as the human eye perceives area changes more dramatically.

Key Terms Glossary

• Distribution of a sample: Values a variable takes and their frequencies.

• Frequency: Count of occurrences.

• Relative frequency: Proportion or percentage of total observations.

• Dotplot: Graph plotting individual data points as dots.

• Histogram: Graphical display of numerical data using bins.

• Density plot: Smoothed curve estimating probability density.

• Bar graph (bar chart): Rectangular bars for categorical data.

• Pareto chart: Bar chart ordered by frequency.

• Pie chart: Circular graphic for proportions.

• Typical value (center): Value representing the center of a distribution.

• Variability: Measure of spread or diversity.

• Outlier: Extreme data point far from the rest.

• Symmetric distribution: Left and right sides are mirror images.

• Bell-shaped distribution: Symmetric, bell-like curve.

• Right-skewed distribution: Longer tail to the right.

• Left-skewed distribution: Longer tail to the left.

• Unimodal distribution: Single peak.

• Bimodal distribution: Two peaks.

• Multimodal distribution: More than two peaks.

• Mode (categorical): Most frequent category.

Purpose of Statistical Graphics

• Discover patterns in data (e.g., identifying dominant groups).

• Communicate findings clearly (e.g., using Pareto charts).

The Future of Statistical Graphics

• Modern visualizations allow for interactive, multi-variable displays (e.g., time, rank, genre, and frequency in one graphic).

Additional info: Density plots and advanced visualizations are mentioned for completeness, though not detailed in the original notes. The glossary and examples are expanded for clarity and exam preparation.