Two students draw histograms of the same class journey times. One makes the widest interval the tallest bar because it contains the most journeys. The other draws a narrower interval taller even though it contains fewer journeys. Could they both be right? The answer depends on whether the horizontal intervals have equal widths and what the vertical axis means.
Frequency density is the number of observations divided by the width of their class interval. When histogram intervals have different widths, the height must represent frequency density so that bar area, not bar height, represents frequency. A taller narrow bar can correctly contain fewer observations than a shorter wide bar. The rule is frequency density = frequency ÷ class width, and frequency = density × width.
- Quick check: read the horizontal axis units and calculate every class width before drawing a bar.
- Make the vertical axis frequency density, with units such as students per minute, when class widths differ.
- Hand-check a bar by multiplying height by width to recover its count.
- Did you know? The class with the highest density is not necessarily the class with the most observations.
- Never invent individual values from grouped intervals: partial-class counts require an explicit distribution assumption.
Why the ordinary height-equals-count shortcut fails
A conventional equal-width frequency histogram can use count on the vertical axis because every bar occupies the same horizontal width; comparing heights then compares counts. If one interval spans five minutes and another spans twenty minutes, a height of twenty on each would give the wide interval four times the rectangular area despite identical printed heights. That picture no longer uses area consistently.
Think of the histogram as assigning a quantity of inked area to each group. A group of 30 journeys must occupy 30 area-units on a frequency-density plot. If the group spans ten minutes, its bar is three density units high; if the same 30 journeys are assigned to a thirty-minute interval, its bar is only one unit high. Width and height trade off so the total area continues to show the same count.
The classroom misconception is often a missing question, not a weak multiplication skill. Before asking ‘Which bar is tallest?’, ask ‘What quantity does this axis measure?’ A frequency histogram and a density-scaled histogram are not interchangeable when the widths vary.
A complete, hand-checked example: 93 invented school journeys
Imagine an illustrative transport survey with 93 journey-time observations. This is made-up practice data, not a survey of actual eduKate students, schools or commuting conditions. We use intervals that include the left endpoint and exclude the right: 0 ≤ t < 5, 5 ≤ t < 15, 15 ≤ t < 35 and 35 ≤ t < 50 minutes. Any recorded endpoint belongs to exactly one class.
- 0 to under 5 min: width 5 min; frequency 8; density 8 ÷ 5 = 1.6 journeys/min.
- 5 to under 15 min: width 10 min; frequency 30; density 30 ÷ 10 = 3 journeys/min.
- 15 to under 35 min: width 20 min; frequency 40; density 40 ÷ 20 = 2 journeys/min.
- 35 to under 50 min: width 15 min; frequency 15; density 15 ÷ 15 = 1 journey/min.
Check the count and the geometry independently
The total frequency is 8 + 30 + 40 + 15 = 93 journeys. The four bar areas are respectively 5 × 1.6 = 8, 10 × 3 = 30, 20 × 2 = 40 and 15 × 1 = 15. Their areas also sum to 93. Both calculations agree, giving us a practical error check.
The highest bar is the 5–15 minute class, because its density is 3 per minute. The largest total class count is 15–35 minutes, because that class contains 40 journeys even though its density is only 2 per minute. Reporting ‘the tallest bar contains the most students’ would be false for this particular histogram.
How to recover an unknown class frequency
Suppose a worksheet supplies only a histogram. The bar from 15 to 35 minutes is drawn 2 density units high. Width = 35 − 15 = 20 minutes. Therefore frequency = 20 min × 2 journeys/min = 40 journeys. The minutes cancel. A learner who produces 2 journeys has mistakenly read density as frequency.
Why a partial-class answer is an estimate
Someone asks, ‘How many took less than ten minutes?’ All eight journeys under five minutes qualify. Within the 5–15 minute class, however, the raw data are unavailable. If we assume observations are evenly distributed within that interval, the five-minute slice from 5 to 10 covers half the interval and contributes approximately 30 × (5 ÷ 10) = 15. So an estimated 23 journeys took less than ten minutes. We do not know that exactly 23 did: all 30 values in the 5–15 class could in principle lie below ten or above it.
This distinction between what the graph guarantees and what an estimate assumes is the core educational skill. A formula can be executed perfectly while the inference is still overstated.
The learner diagnostic: four questions before correction
- Can the learner subtract endpoints to find widths of 5, 10, 20 and 15 minutes? If not, repair interval arithmetic and boundary reading first.
- Can the learner identify a frequency count and a frequency density as different quantities? If not, use units and rectangles.
- Can the learner recover one count by multiplying density × width? If not, draw a rectangle and label each side.
- Can the learner explain why the highest-density class need not contain the most journeys? If not, compare the 5–15 and 15–35 classes aloud.
Give each learner an opportunity to say why a step works. ‘Because that’s the formula’ is weaker evidence than ‘because width × height is area, and area was chosen to represent the number in the interval’. Ask the student to point to a single bar and interpret both axes with units.
Progressive practice: from guided work to independent transfer
Level one: guided, no calculator needed
A bar covers 0–4 hours and has frequency 12. Divide 12 by 4 to obtain height 3 observations per hour. Another covers 4–10 hours and also has frequency 12. Its width is six hours; draw it at height 2 per hour. Although each class has 12 observations, the bar heights are different. Explain that the areas are both 12.
Level two: reconstruct missing values
An interval 10 ≤ x < 18 metres is shown at density 2.5 objects/metre. The width is 8 metres. Calculate frequency = 8 × 2.5 = 20 objects. Now an interval 18 ≤ x < 30 metres contains 24 objects: width 12 metres and density 2 objects/metre. Which bar is taller? The first, despite having fewer objects.
Level three: independent comparison
Design an original two-class survey where a shorter bar represents more observations. One valid answer: class A width 4 and frequency 16 has density 4, whereas class B width 12 and frequency 30 has density 2.5. Bar B is shorter than A but represents 30 rather than 16 counts. The learner should construct a different example and defend it with both area calculations.
Level four: critique an authentic-looking claim
A fictional student reporter writes, ‘The 20–40-minute interval is the busiest because its bar is widest.’ Ask the learner to request the bar height before accepting that inference, then ask for the original frequencies before claiming which exact minute had the most observations. Width alone is about the chosen class boundaries; it does not reveal the distribution inside the class.
Misconceptions worth repairing early
- ‘Higher means more.’ Sometimes it means more per unit of horizontal width. Read the axis before interpreting.
- ‘The widest bar must have more.’ A wide interval can contain few observations; area combines width and height.
- ‘The modal class is the class with the biggest count.’ For unequal widths, the peak density and the largest grouped count can differ; state which comparison you intend.
- ‘A continuous variable guarantees the data are evenly spread inside a class.’ Grouping hides the positions; even spreading is only a model for certain estimates.
- ‘A frequency density of 2.5 means half a person.’ Density is a ratio, not a fractional individual; multiplying by the class width recovers a whole count in a correctly supplied table.
- ‘All histograms have density on the axis.’ Some use counts or relative frequencies with equal bins, and some use normalized probability densities. Inspect labels and conventions.
For parents, students and tutors: what to do next
Students should draw one histogram from a frequency table and one frequency table from a histogram. The reversal is useful: the first tests calculation and scaling, while the second tests whether the chart is genuinely understood. Write units at each step, and verify that every bar area reproduces the original count.
Parents can use a simple household example such as reading-time intervals, but should avoid claiming the measured times came from a school if they did not. Ask, ‘What does this tall bar actually mean?’ rather than rushing to the answer key. A small discussion about what the axis labels mean can reveal more than ten repetitive questions.
Teachers can put two nearby bars of different widths on the same slide and invite a prediction before revealing the count calculation. Tutors should keep an error log distinguishing width errors, density errors, area errors, endpoint assignment errors and unsupported assumptions. Observable progress is the student’s ability to interpret an unfamiliar graph and justify each calculation independently, not a promised test score.
Evidence, context and limits of the method
The formulas follow directly from the geometry of rectangles. Penn State’s Introduction to Probability Theory lesson explains a normalized density histogram where bar area gives relative frequency; multiply those normalized heights by the sample size to recover the count-density convention used in our example. The wider principle is the same: the histogram’s area carries the quantity.
Histograms summarize; they do not recreate raw observations. Different boundary choices can change the displayed pattern. When there are very few observations or precise cutoffs matter, show the original data or a dot plot alongside the histogram. Treat a comparison of density as a property of the chosen intervals, not proof that a particular point inside the interval is the most common.
Frequently asked questions
Does every bar need the same width?
No. Unequal widths are allowed when the vertical scale is chosen correctly. With count-density bars, height × width must give the observation count for each class.
Are the units really observations per minute?
In the journey-time example, yes: frequency is journeys (a count), class width is minutes, and density is journeys/minute. In an ordinary mathematical histogram, units may be omitted from the density label; including them helps beginners reason.
How do I check the entire histogram?
Multiply the height and width of every bar and add the recovered frequencies. The total should match the sample size. Also check there is no overlap or unintended gap between consecutive class boundaries.
Can I know the exact mean from the histogram?
Usually not without raw values. You can calculate a grouped estimate using class midpoints and counts, but that requires treating the midpoint as a representative value, which may not match the actual distribution.
Is the tallest bar always the modal class?
The tallest bar identifies the interval with greatest frequency density. If you mean the interval containing the greatest number of recorded observations, compute the class frequencies instead. State which meaning you are using.
Where to read next
For another skill in reading a histogram honestly, see eduKate’s verified guide to zero-frequency gaps. Continue with the Education Hub and the Clementi explanation-first Mathematics reference, which emphasises identifying the hidden misunderstanding before increasing practice.
A histogram becomes readable when the learner stops treating its bars as pictures and starts treating them as quantities. Check the axis, calculate the width, multiply to verify the area, and say exactly what the data can—or cannot—support.