Displaying and Describing Data

Three Rules of Data Analysis

Effective data analysis begins with visual representation. Making a picture of the data helps clarify patterns and relationships, highlights important features, and communicates findings to others.

• Rule 1: Make a picture to think clearly about patterns and relationships.

• Rule 2: Make a picture to show important features of the data.

• Rule 3: Make a picture to tell others about the data.

The Area Principle

The Area Principle states that the area occupied by a part of a graph should correspond to the magnitude of the value it represents. This principle ensures accurate visual representation and prevents misleading interpretations.

• Bars in bar charts should have equal widths.

• Be cautious with two-dimensional images for one-dimensional data.

Summarizing and Displaying a Categorical Variable

Frequency Tables

A frequency table lists each distinct outcome and its frequency. When there are many outcomes, combining them into broader categories is recommended.

• Definition: Table with outcomes and their counts.

• Application: Useful for summarizing categorical data.

Relative Frequency Tables

A relative frequency table displays each outcome and its relative frequency (percentage of total).

• Definition: Table with outcomes and their relative frequencies.

• Application: Useful for comparing proportions across categories.

Bar Charts

Bar charts visually display the frequency or relative frequency of each category. All bars must have the same width, making them suitable for general audiences.

• Frequency Bar Chart: Shows counts.

• Relative Frequency Bar Chart: Shows percentages.

Pie Charts

Pie charts represent each category as a slice of a circle, with the size proportional to the whole.

• Application: Good for showing proportions in categorical data.

Ring Charts

Ring charts partition a ring into areas proportional to each category's value. They are visually similar to pie charts and may be easier to read.

• Application: Useful for general audiences.

Choosing the Right Chart

Choose the chart that best tells the story of your data, considering the intended audience and the honest expression of data features.

Randomness and Signal Detection

Random Matters

Statistics helps distinguish between random variation (noise) and systematic effects (signal). For example, the distribution of Titanic survivors was not random, as women and children were prioritized, and ticket class influenced survival.

Random Sampling and Visualization

Random selection can be visualized by repeatedly sampling and plotting distributions, such as pie charts for ticket class among survivors. Comparing these to the actual distribution can reveal systematic effects.

Displaying a Quantitative Variable

Histograms

A histogram displays quantitative data, showing the distribution of values. It is useful for identifying the most common values, gaps, and the overall shape.

• Bin Width: The choice of bin width affects the story told by the histogram.

• Gaps: Indicate ranges with no data.

Statistical Software Example

Statistical software (e.g., StatCrunch) can be used to create histograms and summary statistics by selecting columns and options.

Choosing the Right Display

Bar and pie charts are for categorical data, while histograms are for quantitative data. Always consider the variable type before choosing a display.

Shape of Distributions

Modes

The mode of a histogram is a peak or high-frequency bin. Distributions can be unimodal (one mode), bimodal (two modes), or multimodal (three or more modes).

• Unimodal: One peak.

• Bimodal: Two peaks.

• Multimodal: Three or more peaks.

Uniform Distributions

A uniform distribution has bins with equal or nearly equal frequencies, resulting in a flat histogram.

Symmetry

A symmetric distribution looks the same on both sides of its center. Symmetry does not necessarily mean bell-shaped.

Skewness

A histogram is skewed right if the longer tail is on the right, and skewed left if the longer tail is on the left.

• Skewed Right: Tail extends to the right.

• Skewed Left: Tail extends to the left.

Outliers

An outlier is a data value far above or below the rest. Outliers may indicate errors or important values (e.g., CEO income, extreme temperatures).

Example: Interpreting Shape

Consider a histogram of credit card expenditures. The distribution is unimodal, skewed right, and contains an outlier.

Center of a Distribution

The Median

The median is the center value of a data set. Half the values are below and half are above the median. For symmetric distributions, the median is at the center.

• Odd sample size: Median is the middle value.

• Even sample size: Median is the average of the two middle values.

The Mean

The mean is the arithmetic average, calculated as the sum of all values divided by the number of values.

• Formula:

Mean or Median?

Both mean and median are measures of center. Outliers can affect the mean more than the median. The choice depends on the data's shape and the information needed.

• Calculate both and investigate outliers.

• Report the measure most appropriate for the context.

Spread of a Distribution

Range

The range is the difference between the maximum and minimum values. It is sensitive to outliers.

• Formula:

Percentiles and Quartiles

Percentiles divide data into 100 groups. The first quartile (Q1) is the 25th percentile, the median is the 50th percentile, and the third quartile (Q3) is the 75th percentile.

Interquartile Range (IQR)

The interquartile range (IQR) is the difference between Q3 and Q1, measuring the spread of the middle half of the data.

• Formula:

Benefits and Drawbacks of IQR

• IQR is not sensitive to outliers.

• Provides a summary of spread around the median.

• May not be familiar to general audiences.

Standard Deviation and Variance

The variance measures how far data values are spread from the mean. The standard deviation is the square root of the variance and has the same units as the original data.

• Variance formula:

• Standard deviation formula:

Standard Deviation and Histograms

Standard deviation reflects the average distance of values from the mean. Histograms with values close to the mean have small standard deviations; those with values far from the mean have large standard deviations.

Summary: What to Tell About a Quantitative Variable

• Make a histogram.

• Discuss shape: unimodal, symmetric, outliers.

• Describe center and spread: median with IQR, mean with standard deviation.

• Report unusual features: multiple modes, outliers.

Randomness in Sampling

Sample Means and Population Means

Sample statistics vary from sample to sample. The mean of a sample may differ from the population mean, but most sample means are close to each other and provide a good estimate.

Common Pitfalls in Data Display and Summary

• Do not violate the area principle.

• Keep displays honest and accurate.

• Do not use histograms for categorical variables.

• Do not compute numerical summaries for categorical data.

• Choose appropriate bin widths for histograms.

• Sort values before finding the median or percentiles.

• Do not round in the middle of calculations.

• Beware of multiple modes and outliers.

• Use appropriate summaries for the data's shape.

Chapter Review

• Make and interpret frequency tables for categorical variables.

• Display categorical data with bar charts or pie charts using the area principle.

• Display quantitative data with histograms, stem-and-leaf displays, or dotplots.

• Describe distributions by shape, center, spread, and unusual features.

• Compute mean, median, standard deviation, and IQR, and know when to use each.