Representing a Quantitative Variable with Graphs - AP Statistics

SECTION 1

Historical Context & Motivation

Long before the era of statistical software, scientists and social reformers wrestled with the challenge of communicating quantitative patterns to audiences who could not easily parse tables of raw numbers. The impulse to represent quantitative data visually grew from a practical need: the human eye detects shape, spread, and outliers in a graphic far more efficiently than in columns of digits. This historical arc—from hand-drawn bar charts to interactive dashboards—has shaped every graphical tool you will encounter on the AP Statistics exam.

1786 • William Playfair's Bar and Line Charts

Scottish engineer William Playfair published The Commercial and Political Atlas, introducing the bar chart and time-series line graph as tools for economic analysis—the first systematic use of graphical displays for quantitative data.

1832 • André-Michel Guerry's Thematic Maps

French lawyer Guerry used shaded maps and comparative bar charts to display crime rates across French departments, pioneering the idea that distributional shape reveals sociological insight.

1891 • Karl Pearson's Histogram

Pearson coined the term histogram while studying biological variation, formalizing the idea of grouping continuous data into intervals and plotting frequency as area.

1977 • John Tukey's Stem-and-Leaf & Boxplot

In Exploratory Data Analysis, Tukey introduced the stem-and-leaf plot and the box-and-whisker plot, shifting statistical practice toward visual exploration before formal inference.

2000s • Digital Visualization & AP Statistics Curriculum

Software tools like TI calculators, JMP, and R democratized graphical production, and the AP Statistics course made graphical literacy a core competency—not just a supplement to numerical summaries.

The central question driving this lesson is deceptively simple: given a set of quantitative measurements, which graphical display best reveals the shape, center, spread, and unusual features of the distribution? Answering that question well is one of the most frequently tested skills in AP Statistics, and it requires understanding the construction, interpretation, and limitations of dotplots, histograms, stemplots, and boxplots.

SECTION 2

Core Principles & Definitions

Before constructing any graph, it helps to internalize several foundational ideas that govern how statisticians think about displaying quantitative data. A quantitative variable records numerical measurements or counts for which arithmetic operations such as averaging make sense—height in centimeters, reaction time in milliseconds, annual income in dollars. The distribution of a quantitative variable describes the pattern of variation: which values occur, how often, and how they cluster or spread. Every graphical tool in this lesson is a different lens for viewing that distribution.

Shape

A distribution's shape is classified by its symmetry (symmetric, left-skewed, right-skewed), modality (unimodal, bimodal, multimodal), and uniformity. Shape dictates which summary statistics are most appropriate.

Center

The center is a single value that represents a typical observation. Common measures include the mean (balance point) and median (middle value). A good graph makes the center visually apparent.

Spread

Spread (variability) measures how far values deviate from center. Graphs reveal spread through the overall width of the distribution; numerical companions include range, IQR, and standard deviation.

Unusual Features

Outliers, gaps, and clusters are crucial contextual features. A graph should make outliers (observations notably removed from the pattern) immediately visible so the analyst can investigate their cause.

Context

Every graph must be interpreted in context—with reference to the variable measured, its units, and the population or sample. On the AP exam, failing to reference context in free-response answers costs points.

✦

KEY TAKEAWAY

Think of a graphical display as a photograph of a crowd taken from a helicopter. A dotplot is like a close-up photo where you can identify every individual; a histogram is a wide-angle shot that reveals the crowd's overall shape; a boxplot is a thermal-imaging overlay that highlights just the structural landmarks (quartiles and outliers). Each 'camera setting' sacrifices some detail to emphasize different features. Your job on the AP exam is to choose the right setting for the question being asked and to describe what the picture shows using the language of shape, center, spread, and unusual features.

SECTION 3

Visual Explanation — Dotplot, Histogram & Stemplot

The SVG diagram below places three representations of the same data set side by side so you can see how each display emphasizes different aspects of the distribution. The data represent the number of hours 20 students reported studying for an AP Statistics exam: 2, 3, 3, 4, 4, 4, 5, 5, 5, 5, 6, 6, 6, 7, 7, 8, 8, 9, 11, 14.

All three displays depict the same 20 observations. The dotplot (left) preserves every individual value; the histogram (center) groups values into bins and uses bar height to show frequency; the stemplot (right) splits each observation into a stem and leaf, retaining the original digits while revealing shape. Red elements indicate the right-tail values (11 and 14) that may be outliers.

Notice how the three displays convey the same fundamental story—the distribution is roughly unimodal and right-skewed with most students studying between 4 and 8 hours—but each adds its own nuance. The dotplot makes it easy to count exact frequencies: exactly four students studied 5 hours. The histogram trades that precision for a cleaner overall silhouette, which is advantageous when n is large. The stemplot provides a compromise, preserving every digit while arranging data into rows that mimic a sideways histogram. On the AP exam, any of these three may be appropriate for describing a distribution, and you should be comfortable constructing and reading all three.

💡

AP Exam Tip

When asked to "describe the distribution" on a free-response question, always address Shape, Center, Spread, and any unusual features (outliers, gaps, clusters)—all in context. A convenient mnemonic is SOCS (Shape, Outliers, Center, Spread).

SECTION 4

Mathematical Framework — Building Histograms & Boxplots

Although constructing a dotplot or stemplot is largely mechanical, histograms and boxplots rely on quantitative decisions—bin width selection and quartile computation—that you should understand precisely. Mastering these calculations ensures you can both create and critique these displays.

Histogram Construction

BIN WIDTH GUIDELINE

Approximate number of bins ≈ √n

For n = 20, √20 ≈ 4.5, suggesting 4–5 bins. This is a rough starting point; the final choice should produce a display whose shape is clear and stable.

Each bin of a histogram spans a range of values, and the frequency (count) or relative frequency (proportion) of observations falling in that range determines the bar's height. When bins have equal width—the standard in AP Statistics—the height of each bar is directly proportional to the count. If bins differ in width, area rather than height represents frequency, but unequal-width histograms are uncommon on the AP exam. A critical convention is how boundaries are handled: typically, an observation equal to a boundary value is placed in the bin to the right (i.e., bins are left-inclusive: [a, b)).

Five-Number Summary & Boxplot Construction

FIVE-NUMBER SUMMARY

Min, Q₁, Median (Q₂), Q₃, Max

The median is the middle value when data are sorted. Q₁ is the median of the lower half; Q₃ is the median of the upper half.

INTERQUARTILE RANGE

IQR = Q₃ − Q₁

The IQR captures the middle 50% of the data and is used to identify outliers.

OUTLIER FENCES (1.5 × IQR RULE)

Lower fence = Q₁ − 1.5 × IQR Upper fence = Q₃ + 1.5 × IQR

Any observation below the lower fence or above the upper fence is marked as an outlier and plotted individually beyond the boxplot's whiskers.

A modified boxplot (the version expected on the AP exam) extends whiskers only to the most extreme non-outlier data values, not to the fences themselves. Points beyond the fences are plotted as individual dots or asterisks. This distinction is important: the whisker endpoint is an actual data value, not the fence.

SECTION 5

The Modified Boxplot — Anatomy & Interpretation

The boxplot is the only standard AP Statistics graph that does not show frequency directly; instead, it maps the five-number summary and highlights outliers. This makes it especially powerful for comparing distributions across groups. The diagram below labels every component of a modified boxplot for the study-hours data set from Section 3.

A fully annotated modified boxplot of the study-hours data. The pink median line at 5.5 sits left of center within the box, confirming right skewness. The amber box spans Q₁ = 4 to Q₃ = 7.5 (IQR = 3.5). The right whisker extends to 11, the largest non-outlier value; 14 is flagged as an outlier because it exceeds the upper fence of 12.75.

Several interpretive points deserve emphasis. First, the box always contains the middle 50% of the data (the interquartile range), so a wider box signals greater variability in the central half. Second, the position of the median line within the box indicates skewness: a median line closer to Q₁ suggests right skew, while one closer to Q₃ suggests left skew. Third, the relative lengths of the two whiskers reinforce that skewness assessment—the right whisker here is markedly longer. Finally, note what a boxplot does not show: it reveals nothing about modality. A bimodal distribution and a unimodal distribution with the same five-number summary would produce identical boxplots, which is why boxplots are best used for comparison rather than as the sole exploratory display.

SECTION 6

Worked Example — From Raw Data to Graphical Description

Suppose you are given the following commute times (in minutes) for 15 employees at a small firm: 8, 12, 15, 15, 18, 20, 22, 24, 25, 28, 30, 32, 35, 42, 55. Construct a histogram and describe the distribution completely.

Constructing and Interpreting a Histogram

1
Step 1 — Organize the Data

The data are already sorted in ascending order from 8 to 55. There are n = 15 observations, and the range is 55 − 8 = 47 minutes.
2
Step 2 — Choose Bin Width

Using the guideline √15 ≈ 3.9, we aim for about 4 bins. With a range of 47, a bin width of 10 minutes gives 5 bins: [5, 15), [15, 25), [25, 35), [35, 45), [45, 55]. Five bins is reasonable.

Bin width = 10 minutes; 5 bins from 5 to 55
3
Step 3 — Tally Frequencies

[5, 15): 8, 12 → frequency 2. [15, 25): 15, 15, 18, 20, 22, 24 → frequency 6. [25, 35): 25, 28, 30, 32 → frequency 4. [35, 45): 35, 42 → frequency 2. [45, 55]: 55 → frequency 1.

Frequencies: 2, 6, 4, 2, 1
4
Step 4 — Draw the Histogram

On the horizontal axis, mark the bin boundaries (5, 15, 25, 35, 45, 55). On the vertical axis, mark frequencies from 0 to at least 6. Draw bars with heights 2, 6, 4, 2, 1 for the respective bins. Bars should touch (no gaps), reflecting the continuous nature of the variable.
5
Step 5 — Describe the Distribution (SOCS)

Shape: The distribution of commute times is roughly unimodal and skewed to the right, with the peak in the 15–25-minute bin and a long right tail. Outliers: The observation at 55 minutes appears unusually high and is a potential outlier. Center: The median commute time is 24 minutes (the 8th value in the sorted list), suggesting a typical employee commutes about 24 minutes. Spread: Commute times range from 8 to 55 minutes, a span of 47 minutes.

The distribution of commute times is unimodal and right-skewed with a center around 24 minutes, a range of 47 minutes, and a potential outlier at 55 minutes.

⚠

Scoring Note

On AP free-response questions, many students lose credit by omitting context. 'The distribution is skewed right' is incomplete; you must say 'the distribution of commute times is skewed right.' Similarly, a center should include units: 'about 24 minutes,' not just '24.'

SECTION 7

Comparing Graph Types — Strengths & Limitations

No single graph type is universally best; each has trade-offs that make it more or less suitable depending on the data set's size, the question being asked, and the audience for the display. The table below provides a systematic comparison of the four major graph types for quantitative data on the AP Statistics exam.

Feature	Dotplot	Stemplot	Histogram	Boxplot
Preserves individual values	Yes	Yes	No (grouped)	No (summary only)
Shows shape clearly	Yes (small n)	Yes (small n)	Yes (any n)	Partially (skewness only)
Reveals modality	Yes	Yes	Yes	No
Identifies outliers	Visual inspection	Visual inspection	Visual inspection	Formally (1.5 × IQR rule)
Useful for large n	Poor (overcrowded)	Poor (too many rows)	Excellent	Excellent
Best for group comparison	Possible (stacked)	Back-to-back for 2 groups	Side-by-side	Excellent (parallel boxplots)

Comparison of four graphical displays for quantitative data.

✦

KEY TAKEAWAY

Think of each graph type like a different imaging technology in medicine. A dotplot is like an X-ray: quick and revealing for simple cases but overwhelmed by complexity. A histogram is like a CT scan: it bins slices of data to reveal internal structure at any scale. A boxplot is like a vital-signs summary—blood pressure, heart rate, temperature—it boils the patient down to key numbers for rapid triage and comparison. The best diagnostic workup uses multiple views, and the same is true in statistics: pair a boxplot with a histogram or dotplot whenever possible.

SECTION 8

Connection to Inference & Advanced Displays

The graphical tools introduced in this lesson are not merely descriptive—they play essential roles later in the AP Statistics course. When you learn about sampling distributions and inference, you will need to assess normality conditions by examining histograms or dotplots of sample data. A boxplot's outlier identification feeds directly into decisions about whether to use robust statistics (median, IQR) or mean-based statistics (mean, standard deviation). Furthermore, graphical analysis of residuals in regression relies on the same shape-assessment skills you develop here.

This Lesson	Later in AP Statistics
Describing shape (symmetric, skewed)	Assessing normality for t-procedures and z-procedures
Identifying outliers via 1.5 × IQR	Detecting influential points in regression; choosing resistant summaries
Comparing distributions with parallel boxplots	Two-sample inference; ANOVA motivation
Histogram of raw data	Histogram of sampling distribution to illustrate the Central Limit Theorem
Stemplot for small data sets	Normal probability plots (Q-Q plots) in advanced courses extend the idea of ordering data

How graphical skills connect to later AP Statistics topics.

Beyond the AP exam, modern data science employs density plots (smooth curves fitted over histograms), violin plots (which embed a density curve inside a boxplot outline), and interactive dashboards that let users change bin widths dynamically. These tools all build on the same foundational idea: a well-chosen graphical display reveals distributional features that summary statistics alone cannot capture. Mastering the simpler displays now gives you the conceptual scaffolding for every sophisticated visualization you will encounter in college-level statistics and data science.

SECTION 9

Practice Problems

Question 1: CONCEPTUAL

A researcher collects the ages of 200 participants in a clinical trial and wants to display the distribution's overall shape. Which graph would be most appropriate?

A. Dotplot
B. Stemplot
C. Histogram
D. Pie chart
E. Bar chart

Show answer (C)

C — Histogram. With n = 200, a dotplot (A) would be overcrowded with stacked dots, and a stemplot (B) would produce an impractically long display—both are better suited for small data sets. A pie chart (D) is used for categorical data proportions, not quantitative distributions. A bar chart (E) is also for categorical data (the bars represent categories, not numerical intervals). A histogram groups continuous quantitative data into bins and uses bar height for frequency, making it the standard choice for displaying the shape of a distribution when the sample size is moderately large or large.

Question 2: BASIC CALCULATION

For the data set {3, 7, 8, 10, 12, 15, 18, 22, 25, 30}, Q₁ = 8 and Q₃ = 22. Using the 1.5 × IQR rule, which values would be classified as outliers?

A. 3 only
B. 30 only
C. 3 and 30
D. No values are outliers
E. 7 and 25

Show answer (D)

D — No values are outliers. IQR = Q₃ − Q₁ = 22 − 8 = 14. Lower fence = Q₁ − 1.5 × IQR = 8 − 21 = −13. Upper fence = Q₃ + 1.5 × IQR = 22 + 21 = 43. An outlier must fall below −13 or above 43. The minimum value (3) is well above −13, and the maximum value (30) is well below 43, so no observations qualify as outliers. Choice A is wrong because 3 > −13. Choice B is wrong because 30 < 43. Choice C combines both errors. Choice E identifies inner values with no outlier justification.

Question 3: INTERMEDIATE

A histogram of test scores is strongly skewed left with a single peak around 88 out of 100. Which statement best describes the relationship between the mean and median of this distribution?

A. The mean is approximately equal to the median.
B. The mean is greater than the median.
C. The mean is less than the median.
D. The mean and median are both equal to 88.
E. The relationship cannot be determined from the histogram alone.

Show answer (C)

C — The mean is less than the median. In a left-skewed distribution, the long tail of low values pulls the mean toward the left (toward lower values) while the median, being the positional middle, is more resistant to those extreme low scores. Therefore the mean < median. Choice A describes a symmetric distribution. Choice B describes a right-skewed distribution (the opposite situation). Choice D is too specific and assumes the mode equals both measures, which is not generally true. Choice E is incorrect because the direction of skewness reliably indicates the mean-median relationship for unimodal distributions.

Question 4: APPLIED

4 pts

A quality engineer measures the fill volumes (in mL) of 25 bottles from a production line. The sorted data are: 498, 499, 499, 500, 500, 500, 501, 501, 501, 501, 502, 502, 502, 502, 502, 503, 503, 503, 504, 504, 505, 505, 506, 508, 515. (a) Calculate the five-number summary. (b) Determine whether there are any outliers using the 1.5 × IQR rule. (c) Sketch a modified boxplot (describe key features if you cannot draw). (d) The engineer claims the filling process is stable and centered at 502 mL. Using your graphical analysis, evaluate this claim.

Show answer

Rubric (4 points): Point 1 — Five-number summary (Part a): Min = 498. With n = 25, the median is the 13th value = 502. The lower half consists of positions 1 through 12; the median of these 12 values is the average of positions 6 and 7: (500 + 501)/2 = 500.5, so Q1 = 500.5. The upper half consists of positions 14 through 25; the median of these 12 values is the average of positions 19 and 20: (504 + 504)/2 = 504, so Q3 = 504. Max = 515. Five-number summary: 498, 500.5, 502, 504, 515. Point 2 — Outlier identification (Part b): IQR = Q3 minus Q1 = 504 minus 500.5 = 3.5. Lower fence = 500.5 minus 1.5 times 3.5 = 500.5 minus 5.25 = 495.25. Upper fence = 504 plus 1.5 times 3.5 = 504 plus 5.25 = 509.25. The minimum value of 498 is above 495.25, so there are no low outliers. The maximum value of 515 exceeds 509.25, so 515 mL is an outlier. Point 3 — Modified boxplot description (Part c): The box extends from Q1 = 500.5 to Q3 = 504 with a median line at 502. The left whisker extends from Q1 down to 498, which is the minimum and is not an outlier. The right whisker extends from Q3 up to 508, which is the largest data value at or below the upper fence of 509.25. The value 515 is plotted as an individual point beyond the right whisker. The median is roughly centered in the box, but the right whisker is longer than the left whisker, suggesting slight right skew driven by the outlier. Point 4 — Evaluation of claim (Part d): The median of 502 mL supports the claim that the process is centered at 502. However, the presence of the outlier at 515 mL, a bottle overfilled by 13 mL, suggests the process may not be fully stable. This single extreme value could indicate an intermittent equipment malfunction. While the bulk of the data (the middle 50 percent, from 500.5 to 504) is tightly clustered around 502, the engineer should investigate the cause of the 515 mL reading before concluding stability.

Question 5: CRITICAL THINKING

4 pts

Two AP Statistics students each display the same data set—the 50-meter dash times (in seconds) for 30 track athletes—using different histograms. Student A uses a bin width of 0.5 seconds and obtains a bimodal distribution. Student B uses a bin width of 2 seconds and obtains a unimodal, roughly symmetric distribution. (a) Explain how the choice of bin width can produce these two different appearances from the same data. (b) Which student's histogram is more likely to reveal a meaningful feature of the data? Justify your answer. (c) Suggest a strategy for choosing an appropriate bin width and explain how using an additional type of graph could resolve the apparent contradiction. (d) A third student creates a boxplot of the same data and reports that the distribution appears symmetric with no outliers. Explain why this finding does not contradict Student A's bimodal histogram.

Show answer

Rubric (4 points): Point 1 — Explanation of bin width effect (Part a): A histogram's appearance depends critically on bin width. A narrow bin width (0.5 s) provides higher resolution, revealing local peaks and valleys in the frequency distribution; if the data have two clusters of times (e.g., one group around 6.5 s and another around 7.5 s), narrow bins can separate them into visible modes. A wide bin width (2 s) smooths over such detail by lumping observations from both clusters into a single bar, making the distribution appear unimodal. The underlying data have not changed—only the level of aggregation. Point 2 — Which is more informative (Part b): Student A's bimodal histogram is more likely to reveal a meaningful feature. Bimodality often reflects an underlying categorical grouping—for example, the 30 athletes might include both sprinters and middle-distance runners, or male and female athletes, with genuinely different typical dash times. A unimodal display that hides this structure would miss an important aspect of the data. Student B's wide bins have over-smoothed the data, concealing real structure. Point 3 — Strategy for bin width and additional graph (Part c): A sound strategy is to try several bin widths and look for features that persist across multiple choices—a mode that appears under multiple bin widths is likely real, while one that vanishes with a slight change may be an artifact. Additionally, constructing a dotplot of the same data (feasible since n = 30 is small enough) would show every individual value and make any clustering immediately visible without the ambiguity introduced by binning. A stemplot would serve the same purpose. Using a display that preserves individual values alongside a histogram resolves the contradiction by confirming or refuting bimodality. Point 4 — Why the boxplot does not contradict bimodality (Part d): A boxplot displays only the five-number summary (and outliers); it does not convey information about modality. A bimodal distribution with two symmetric clusters can produce a five-number summary in which the median falls between the two modes, Q₁ and Q₃ are roughly equidistant from the median, and the whiskers are approximately equal in length—yielding a perfectly symmetric-looking boxplot. Because the boxplot cannot show the valley between two modes, its appearance of symmetry is entirely consistent with Student A's bimodal histogram. This illustrates a fundamental limitation of boxplots: they should not be used as the sole display when assessing the shape of a distribution.

SUMMARY

Lesson Summary

Summary

Graphical displays are the statistician's first line of inquiry for understanding a quantitative variable's distribution. A dotplot places one dot per observation on a number line, preserving every value and working best for small data sets. A stemplot splits each number into a stem and leaf, retaining original digits while producing a sideways histogram shape. A histogram groups data into equal-width bins and uses bar height for frequency—ideal for moderate to large samples but sensitive to bin width choice. A modified boxplot maps the five-number summary (Min, Q₁, Median, Q₃, Max) and flags outliers using the 1.5 × IQR rule, making it the tool of choice for comparing groups via parallel boxplots, though it cannot reveal modality.

When describing any graphical display on the AP exam, always address shape (symmetric, left-skewed, or right-skewed; unimodal or bimodal), center (mean or median with units), spread (range, IQR, or standard deviation with units), and any unusual features such as outliers, gaps, or clusters—always in context. Remember the mnemonic SOCS (Shape, Outliers, Center, Spread), and pair numerical summaries with graphical displays for the most complete analysis.