Summarizing Data — Summarize data with tables and charts

The challenge of summarizing large collections of observations is as old as organized society itself. Ancient civilizations maintained census records and trade inventories, yet the sheer volume of raw numbers made it nearly impossible to extract meaningful insights without some form of condensation. The modern practice of data summarization — organizing data into frequency tables, bar charts, histograms, and pie charts — emerged gradually through centuries of innovation in both tabular record-keeping and graphical representation. Understanding this history reveals why certain conventions exist and why thoughtful summarization remains a cornerstone of statistical reasoning.

Each of these milestones addressed a fundamental question: how can we condense a potentially enormous data set into a compact representation that preserves the essential structure — the center, spread, shape, and notable features — while remaining interpretable at a glance? This question motivates the techniques developed throughout this lesson, from simple frequency tables through the rich visual vocabulary of modern charts.

Before constructing any table or chart, it is important to recognize the foundational ideas that govern effective data summarization. The goal is not merely to shrink a data set but to organize it so that the underlying distribution — the pattern of how values occur — becomes transparent. The following principles guide every decision, from choosing class widths in a histogram to selecting chart types for a presentation.

The diagram below illustrates the complete summarization pipeline for a categorical data set. On the left, raw survey responses about students' preferred study locations are listed. In the center, those responses have been tallied into a frequency table showing each category's count and relative frequency. On the right, the frequencies are rendered as a bar chart, making the dominant category and comparative magnitudes immediately visible.

This three-stage process — raw data, tabulation, and visualization — is the conceptual backbone of every summarization technique we will study. The table provides exact counts for computation, while the chart provides an immediate perceptual comparison of magnitudes. In practice, both representations are complementary: the table supports precise calculations of proportions and totals, whereas the chart communicates the overall pattern to a broader audience.

Although tables and charts are often treated as informal tools, each rests on precise definitions that connect to probability and further statistical analysis. Mastering these definitions ensures that the summaries you construct are mathematically sound and correctly interpreted.

These formulas are interconnected. Relative frequencies serve as empirical estimates of probabilities, cumulative frequencies underpin distribution functions, and class-width decisions directly affect the visual impression a histogram conveys. As a general rule, use too few classes and you obscure structure; use too many and you introduce noise.

Selecting the right chart type is critical: a histogram applied to categorical data or a pie chart used with continuous data will mislead the reader. The following diagram and table classify the major chart types by the type of data they are designed to display and the analytical question they answer.

A common mistake is to use a histogram for categorical data or a bar chart for continuous data. The distinction matters because the touching bars of a histogram encode the fact that the underlying variable has no gaps between adjacent intervals — it is measured on a continuous scale. In a bar chart, the gaps between bars visually reinforce that the categories are discrete and unordered (or ordered only by convention, as with letter grades).

Suppose you have collected the exam scores of 30 students in a finite mathematics course. The raw data are as follows: 52, 58, 61, 63, 65, 67, 68, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 87, 88, 90, 91, 93, 95, 98. We will construct a grouped frequency table with class boundaries, frequencies, relative frequencies, and cumulative frequencies, and then describe the resulting histogram.

No single table or chart is universally optimal. Each tool embodies a trade-off between detail and clarity, and its effectiveness depends on the data type, the audience, and the analytical goal. The table below contrasts the primary strengths and limitations of the most common summarization tools covered in this lesson.

The tables and charts developed in this lesson are not merely presentation devices — they are the empirical foundation upon which more advanced statistical concepts are built. The relative frequency distribution obtained from a large sample approximates the theoretical probability distribution of the underlying random variable, and the histogram converges to the probability density function (pdf) as the sample size grows and the class width shrinks. Similarly, the cumulative relative frequency (ogive) converges to the cumulative distribution function (CDF), which is central to hypothesis testing and confidence intervals.

This forward-looking perspective underscores why getting the descriptive stage right matters so much. If your frequency table has miscounted observations, or your histogram uses class widths that obscure the true shape of the distribution, any subsequent inference — from estimating population parameters to testing hypotheses — will rest on a flawed foundation. The skills you develop in this lesson therefore carry directly into every subsequent chapter of probability and statistics.