Find your next learning step
Choose the question closest to what you need today, or read the guide from the beginning.
A histogram bar can be taller yet represent fewer observations than a shorter bar. The reason is width. In a standard frequency-density histogram, frequency is represented by bar area, not by height alone. A wide low rectangle can have more area than a narrow tall one.
This guide explains that distinction through numbers you can inspect without a complicated diagram. It then shows how to recover frequencies, build a histogram table, compare intervals, and recognise when a grouped display cannot answer an exact question. All datasets are original teaching examples rather than actual school results or measured populations.
Use the 2027 G3 Mathematics syllabus for official curriculum scope and your school’s sequence for timing. This is a G3-focused explanation of histogram reasoning. It is not a claim that every topic here is required in identical depth across G1, G2 and G3.
Contents
Sections 1–5
Sections 6–10
Sections 11–15
Sections 16–20
Sections 21–25
Sections 26–30
Sections 31–31
Suppose one class interval has width five and frequency density six. Its frequency is 5 × 6 = 30. Another has width fifteen and density three. Its frequency is 15 × 3 = 45. The second bar is half as tall but contains fifteen more observations because it is three times as wide.
The relationship is frequency density = frequency ÷ class width. Rearranging gives frequency = frequency density × class width. Before comparing bars, read the horizontal intervals and the vertical-axis label. Do not assume the vertical value is a count merely because the display looks like a familiar bar chart.
When all class widths are equal, comparing heights also compares frequencies because every height is multiplied by the same width. When widths differ, that shortcut is no longer reliable. Always identify what the display encodes before using its most visually striking feature to answer a question.
Frequency is the number of observations in an interval. If eighteen journey times lie between ten and twenty minutes, that interval’s frequency is eighteen. Frequency density describes how much frequency is represented per unit of the measured variable’s scale. Here the class width is ten minutes, so the density is 1.8 observations per minute of interval width.
This does not mean that exactly 1.8 journeys lasted each minute. It is a summary across the whole class. The raw observations could be clustered near one end or spread unevenly. A standard rectangular bar does not preserve those individual positions. Its height represents a class-level concentration measure.
The distinction is similar to comparing the total number of people in two rooms with the number per square metre. One room can have a lower concentration but more people because it is larger. In a histogram, interval width takes the role of size, and area reconnects concentration with total frequency.
Suppose one interval covers five minutes and another covers twenty minutes. Even if observations are similarly concentrated across time, the second interval covers four times as much horizontal territory. A count in that interval cannot be compared fairly with the first count merely by drawing a wider rectangle at the same count height.
If height directly represented frequency while width also represented class width, the rectangle’s area would exaggerate the wider interval’s count. Dividing frequency by width corrects that problem. Multiplying the resulting density by width recovers exactly the original frequency, so the visual area has a consistent meaning.
The purpose is not to make every bar the same height. A genuinely concentrated interval should still appear taller. The adjustment ensures that height describes concentration and area describes count. These are related but different questions, and a good display keeps their meanings consistent across every class.
A vertical axis labelled frequency density calls for a different interpretation from an axis labelled frequency. The words matter as much as the numbers. A density reading of four is not automatically four students. You need the width of that bar’s interval to calculate its frequency.
Also inspect whether the density scale is direct or whether the problem specifies a separate area-to-frequency scale. In a standard direct frequency-density histogram, width times height gives frequency numerically. In a display where one square centimetre represents a stated number of observations, use that stated scale instead of silently assuming a direct density axis.
Do not measure the printed bar with a ruler and treat the page centimetres as the variable’s units. Use the horizontal numerical scale. A class from twenty to thirty-five minutes has width fifteen minutes, regardless of whether the drawing occupies three centimetres or six centimetres on your screen.
A histogram of invented waiting times has a bar for 12 ≤ t < 20 minutes. Its frequency density is 2.5. The class width is 20 − 12 = 8 minutes. The frequency is therefore 8 × 2.5 = 20 observations. The density can be decimal even though the recovered frequency is a whole-number count.
Write the subtraction if the interval boundaries are easy to misread. Using twenty as the width would confuse an upper boundary with an interval length. The interval does not begin at zero. Its width is the difference between its endpoints, not the larger endpoint alone.
Check the units in words: observations per minute multiplied by minutes gives observations. This is a dimensional explanation of why multiplication is appropriate. If you divide density by width instead, the resulting units would not describe a count. Units help distinguish two calculator operations that might both produce plausible-looking decimals.
Suppose twenty-four observations lie in 20 ≤ t < 32 minutes. The class width is twelve minutes. For a direct frequency-density vertical axis, the bar height is 24 ÷ 12 = 2. Draw the rectangle over the full interval from twenty to thirty-two, rising to density two.
The height is not twenty-four. A height of twenty-four over width twelve would represent an area of 288, which is twelve times the intended count. This is the error frequency density is designed to prevent. A wider class needs a lower density for the same frequency because its observations are spread across more interval width.
When constructing several bars, calculate every width before every density. A table with interval, width, frequency and density makes the relationships visible and prevents you from applying one width to all rows by habit. The arithmetic is usually simple; the challenge is preserving what each column represents.
Class A covers 0 ≤ t < 5 and has density six. Class B covers 5 ≤ t < 20 and has density three. A is taller because six exceeds three. B is wider because fifteen exceeds five. To compare their counts, calculate their areas: A has thirty observations and B has forty-five.
Therefore A has the greater frequency density, while B has the greater frequency. Both statements are correct. A question asking which class contains more observations needs the second comparison. A question asking where observations are more concentrated per unit of interval width needs the first.
Avoid declaring one bar “bigger” without saying which feature you mean. It may be taller, wider, greater in area or associated with a different endpoint. Precise language prevents visual impressions from becoming unsupported statistical conclusions. In a written explanation, name frequency or density explicitly rather than using “higher” as a complete answer.
The following original dataset summary describes eighty waiting times. The intervals meet without overlap, and each lower boundary is included while its upper boundary is excluded. The table includes every quantity needed to construct or interpret a standard frequency-density histogram on the stated minute scale.
| Waiting-time interval in minutes | Class width | Frequency | Frequency density |
|---|---|---|---|
| 0 ≤ t < 5 | 5 | 20 | 4 |
| 5 ≤ t < 15 | 10 | 30 | 3 |
| 15 ≤ t < 30 | 15 | 18 | 1.2 |
| 30 ≤ t < 50 | 20 | 12 | 0.6 |
The first bar is tallest, but the second class has the greatest count. The total frequency is 20 + 30 + 18 + 12 = 80. Adding the heights gives 8.8, which is not the total number of observations. Densities cannot be added across unequal widths to recover a total count.
To check the table, multiply each density by its width. The products return twenty, thirty, eighteen and twelve. This row-by-row verification is useful after drawing too: each bar’s numerical area should match the corresponding class frequency. A correct total alone cannot reveal two errors that happen to cancel.
If every class has width ten, a class with frequency twenty has density two and a class with frequency forty has density four. Doubling the height doubles the frequency because the widths are identical. In this special situation, a height comparison happens to give the same ordering as an area comparison.
That success can create a misleading habit. When a later question introduces unequal widths, a student may continue comparing heights without noticing that a condition has changed. The old shortcut was not universally valid; it was valid because a common factor, the width, was shared by every bar.
This is a useful general mathematical lesson. A method may work in several examples because they share a hidden condition. Understanding means noticing the condition and checking whether it survives the next question. Here the question is simple: are the interval widths equal? If not, return to area.
Histograms usually use adjoining bars for adjoining intervals of a numerical variable. The horizontal position and width have quantitative meaning. In an ordinary categorical bar chart, separate categories such as bus, train and walking do not form numerical intervals with meaningful widths on a continuous measurement scale.
Saying only “histogram bars touch” misses the statistical structure. A drawing can have touching rectangles without becoming a correctly constructed histogram. You still need appropriate intervals, a consistent area meaning and a clear vertical scale. Conversely, an empty interval may appear as a gap because its bar has zero height.
Ask what the horizontal dimension represents. If it is time, mass or another numerical measurement, interval width can matter. If it is a list of unrelated categories, bar width is normally a design choice rather than a data value. Read the representation before importing a rule from a different kind of chart.
For adjoining classes 0 ≤ t < 5 and 5 ≤ t < 15, a value of exactly five belongs to the second class. It is not counted twice. The symbols create a non-overlapping assignment. A table that includes five in both neighbouring classes would make the frequencies ambiguous.
When constructing a histogram from rounded measurements, class boundaries may require additional care. If integer-recorded values one through five stand for measurements rounded to the nearest whole unit, a continuous-boundary representation can use 0.5 to 5.5 under the stated rounding model. Do not impose such boundaries if the question uses a different convention.
The histogram’s mathematical meaning depends on the measurement and grouping process. Keep exact values, rounded records and interval definitions distinct. If the question already provides boundaries, use them directly rather than “correcting” them from an unrelated textbook convention. The data contract comes before a familiar-looking layout.
Using the eighty-observation table, the proportion with waiting time below fifteen minutes is (20 + 30) ÷ 80 = 50/80 = 5/8. As a percentage, that is 62.5%. We can add those two full class frequencies because both entire intervals satisfy the requested condition.
The proportion waiting at least thirty minutes is 12/80 = 15%. The proportion from fifteen minutes up to but not including fifty is (18 + 12)/80 = 37.5%. Each calculation uses recovered counts and the overall count, not the sum of bar heights.
Notice that the endpoints in these questions match class boundaries. That makes the exact grouped-frequency answer possible. A threshold inside a class requires extra information or an assumption about distribution within that class. Recognising that difference is an important part of honest statistical reasoning, not a sign that the question has defeated you.
In the table, thirty observations fall between five and fifteen minutes. How many are below ten minutes? The grouped data do not give an exact count within that class. The thirty observations could lie mostly below ten, mostly above ten, or in many other arrangements consistent with the same bar.
If the task explicitly asks for an estimate using an even distribution within the class, half the interval width lies below ten, so estimate half its frequency: fifteen. Add the twenty observations below five to estimate thirty-five below ten. State that this uses an assumption, not a recovered exact count.
The flat top of a histogram rectangle does not prove that the observations are uniformly distributed inside the class. It is a graphical summary chosen to represent the class frequency. Do not silently upgrade a drawing convention into detailed knowledge about every subinterval. An estimate is useful when its basis is visible.
With unequal widths, the tallest frequency-density bar identifies the interval with the greatest density. This is commonly the modal class when interpreting a frequency-density histogram: it indicates the greatest concentration per unit width. In our table, the interval from zero to five minutes has density four, higher than the other bars.
The interval from five to fifteen minutes contains more observations in total, thirty rather than twenty, but covers twice the width. If a question asks for the class with the largest frequency, answer that separate question. Do not use the ambiguous phrase “most common” without checking whether it means concentration or total class count.
For assessment terminology, follow the wording and conventions of the course and question. A safe explanation names both quantities when confusion is possible: “The first interval has the highest frequency density; the second has the highest frequency.” That statement conveys the underlying facts even where an informal label could be misunderstood.
A direct frequency-density histogram has three bars for 0 ≤ x < 4, 4 ≤ x < 10 and 10 ≤ x < 20. Their heights are five, three and two respectively. Find the total number of observations, the percentage in the final interval, and the interval with the greatest concentration.
The widths are four, six and ten. Multiplying by the heights gives frequencies twenty, eighteen and twenty. The total is fifty-eight. The final interval contains 20/58 of the observations, which is approximately 34.5% to one decimal place. Keep the exact fraction until the requested rounding stage.
The first interval has the greatest concentration because its density is five. The first and final intervals have equal frequencies despite different heights and widths. A complete answer distinguishes all three findings rather than using the tallest bar to answer every part. This question deliberately separates density, count and proportion.
Question A: a class from ten to eighteen has density 3.5. Find its frequency. Question B: a class of width twelve contains thirty observations. Find its density. Question C: compare a width-four bar of height seven with a width-ten bar of height three. Which has greater density, and which has greater frequency?
Question D: three adjoining intervals have widths five, ten and fifteen, with densities four, two and two. Find the total frequency. Question E: in D, find the fraction of observations in the last interval. Question F: explain why adding the three heights does not give the total frequency.
Question G: the final interval in D spans fifteen to thirty. Can you state the exact count from fifteen to twenty using the grouped information alone? If not, give an estimate under an explicit uniform-within-class assumption. Keep the exact and estimated parts of your explanation separate.
For A, the width is 18 − 10 = 8, so the frequency is 8 × 3.5 = 28. For B, density is 30 ÷ 12 = 2.5. For C, the first density is greater, seven compared with three, but its frequency is twenty-eight compared with thirty. The second bar contains two more observations.
For D, the frequencies are 5 × 4 = 20, 10 × 2 = 20 and 15 × 2 = 30, giving seventy altogether. For E, the fraction is 30/70 = 3/7. For F, heights describe counts per unit width; their widths differ, so counts require multiplication before addition.
For G, the exact count is unknown from the grouped data. With an assumed even spread, the subinterval from fifteen to twenty occupies five of the class’s fifteen width units, so estimate 5/15 × 30 = 10 observations. The histogram does not establish that ten is the actual count.
Some problems give a drawing without a direct density axis but state that one bar represents a known frequency. Suppose that bar occupies six square centimetres on the printed diagram and represents twenty-four observations. The area scale is four observations per square centimetre, provided the same drawing scale applies throughout.
Another bar occupying nine square centimetres then represents thirty-six observations. This is an area comparison, not a comparison of heights alone. If the original drawing is resized, both measured areas change, so the calibration must be taken from that same resized drawing. Do not combine a ruler measurement from one version with a scale from another.
When the axes already provide numerical density and variable units, use them instead of measuring physical page area unnecessarily. The calibrated-area approach explains why proportional area is the essential feature, while the direct density formula is its convenient numerical form. Always use the scale actually specified by the problem.
Suppose a class from zero to ten has frequency twenty. Splitting it into zero to four and four to ten does not determine the two new frequencies. They must add to twenty, but many pairs are possible. A split of eight and twelve is only one possibility, corresponding to equal density across the original class.
If the raw data are available, recount them into the new intervals. If they are not, state any assumption used to estimate the split. The original histogram cannot reveal individual positions that grouping has discarded. Making the bars narrower on paper does not create new observations or new knowledge.
This connects with the series lesson on grouped tables hiding different exact means. Grouped displays are useful summaries, but they lose detail. Understanding what was retained and what was discarded helps you avoid exact claims that the available information cannot support, even when the diagram looks detailed and numerical.
Combine a width-five interval with frequency twenty and a neighbouring width-ten interval with frequency thirty. The combined width is fifteen and the combined frequency is fifty. The new density is 50 ÷ 15 = 10/3. It is not the simple average of the original densities four and three.
The correct combined density weights each density by its interval width: (4 × 5 + 3 × 10) ÷ 15. The wider interval contributes more horizontal extent, so it receives more weight. A simple average would incorrectly give the two unequal widths equal influence on the combined concentration.
This weighted-average structure appears in many Mathematics topics. Total count divided by total width resembles total distance divided by total time in average speed. The underlying habit is the same: reconstruct compatible totals before forming a combined rate. Avoid averaging rates just because two rate values are printed side by side.
A duration interval from zero to two minutes has width two minutes and contains twenty-four observations. Its density is twelve observations per minute of interval width. Expressed in seconds, the same interval has width 120 seconds, so its density is 24 ÷ 120 = 0.2 observations per second of interval width.
The count remains twenty-four. Multiplying either width by its corresponding density recovers that same frequency. A smaller numerical density does not mean observations disappeared; its unit changed. This is why density comparisons require compatible horizontal units and correctly interpreted vertical labels.
If two graphs use different units, convert before comparing height numbers. Twelve per minute and 0.2 per second are equivalent concentrations. Comparing twelve with 0.2 as bare numbers would ignore their meanings. Units are not only appended to final answers; they help control which quantities can legitimately be compared.
Two classes may have the same interval boundaries but different total numbers of observations. A frequency-density height still reflects count per unit width, so a larger sample can produce taller bars even when its relative distribution is identical. Comparing groups may require relative frequency density rather than raw frequency density.
Suppose Group A has twenty observations in a width-five interval out of one hundred overall. Group B has ten in the same interval out of twenty-five overall. A has greater raw density, four compared with two. But B has the larger proportion in that interval, forty per cent compared with twenty per cent.
Ask what the comparison is intended to establish. “More observations” is a count claim; “a greater share of the group” is a proportional claim. This connects to the series guide on cumulative-frequency comparisons and group totals. A denominator can change a conclusion even when the displayed bar itself has been read correctly.
If a class interval has zero frequency, its density is zero for any positive class width. A histogram may show an empty horizontal stretch there. That gap means the grouped data contain no observations in the interval, not that the scale jumps over it or that its width should be removed.
For example, if intervals cover zero to five, five to ten and ten to twenty, a zero frequency in the middle interval leaves a gap from five to ten. The final bar still begins at ten. Moving it left to close the gap would misrepresent the variable’s scale and the location of observations.
Distinguish zero frequency from missing data. If an interval’s count is unknown, a blank space without explanation should not be interpreted as zero. A well-defined question states the complete table or marks omissions. Mathematics can calculate from known counts; it cannot turn an unreported count into a confirmed absence.
An interval labelled “fifty minutes or more” has no stated finite upper boundary. Its width is not determined, so you cannot calculate an ordinary finite frequency density from its frequency alone. A histogram construction question needs a suitable finite boundary or another explicitly stated convention for displaying that class.
Do not invent an upper boundary of sixty merely because preceding classes happen to have width ten. The open-ended class might include much larger values. Choosing a boundary silently would change the represented interval and could distort the count-per-width interpretation. Ask whether the task supplies additional information.
This is another example of checking whether a formula has all its required inputs. Frequency density needs a frequency and a positive class width. If one is missing, identify the missing information instead of forcing a numerical answer. A clear explanation of insufficiency can be more mathematically sound than an unsupported calculation.
If a student says the tallest bar contains the most observations, ask them to calculate two areas with unequal widths. Keep the numbers easy enough that the comparison is undeniable. The repair is to reconnect height with density and area with count, rather than merely warning them not to trust diagrams.
If they divide frequency by the upper boundary, ask where the class begins. Mark both endpoints and subtract them. If they add densities to find a total, ask what each density’s unit means. Recovering the counts before adding restores a coherent meaning to the operation.
If they state an exact count inside a class, ask whether two different raw datasets could produce the same grouped frequency but different answers to that subinterval question. Inventing such datasets makes information loss visible. Each misconception requires a different correction because each confuses a different part of the representation.
First work from a frequency table to widths and densities. Then reverse the process from widths and densities to frequencies. Finally compare two bars whose heights and counts have opposite orderings. This sequence builds both directions of the relationship and tests whether the learner can use it flexibly.
Add a question with equal widths to explain why the height shortcut works there. Then change just one width and ask what must be recalculated. Contrasts that isolate one changing feature are often more instructive than a long page in which every number and interval changes at once.
Later, include an inside-class estimate and require a sentence naming the uniformity assumption. Statistical literacy includes knowing what the table does not specify. A student who can distinguish an exact recovered count from an assumed estimate is doing more than applying a formula: they are reading the limits of the data.
The broader histogram reading guide provides a route for revisiting the representation. Use this article’s contrasting bars to test the central condition first. If you still treat height as count, return to widths and areas before moving on to more elaborate statistical summaries.
The Mathematics learning hub connects statistics with rates, units and examination interpretation. In particular, unit-rate reasoning explains why dividing a total by a width produces a density. These links are useful because the same mathematical relationship is appearing in different contexts, not because every article must be read in one sitting.
For families exploring teaching support, the Secondary 1 Mathematics Clementi tutorial page describes a foundation-first approach. It is a separate service page, not a statement that this G3 histogram material belongs in every Secondary 1 lesson. Follow the student’s actual subject level, school sequence and current learning need.
Question H: two bars have equal area but widths six and twelve. If the first density is five, find the second density. Question I: a width-eight interval has density 1.5. Express the count and explain why a decimal density does not imply a fractional person.
Question J: a class from twenty to forty contains thirty-two observations. Estimate the count from twenty to twenty-five under an even-spread assumption. Question K: two neighbouring classes of widths four and six have frequencies twelve and twenty-four. Find the density after combining them.
For H, the first frequency is thirty, so the second density is 30 ÷ 12 = 2.5. For I, frequency is twelve; density is a count per unit width, not a headcount. For J, estimate 5/20 × 32 = 8, explicitly conditional on the assumption. For K, combined density is (12 + 24)/(4 + 6) = 3.6.
A histogram can show the distribution of a measured variable in a dataset. It cannot by itself establish why those values occurred. A tall concentration of short waiting times does not prove that a particular staffing policy caused the pattern. The display contains grouped outcomes, not a controlled causal comparison.
Similarly, a bar representing many high scores does not demonstrate which teaching method produced them. Sample selection, prior attainment, measurement conditions and other factors may matter. This guide uses invented values precisely to teach representation without attaching unsupported claims about schools, tutors or student groups to the shape.
Keep descriptive and causal statements separate. “Thirty observations lie in this interval” can be supported by the frequency calculation. “This intervention caused those thirty observations” needs additional evidence and an appropriate design. Accurate histogram arithmetic is valuable, but it does not supply information the dataset and its collection process do not contain.
Suppose a histogram represents ninety observations. Two known bars have widths five and ten and densities four and three. Their frequencies are twenty and thirty, accounting for fifty observations. A third bar has width twenty, and it is the only remaining class. Its frequency must be ninety minus fifty, which is forty. Its density is therefore forty divided by twenty, or two.
This is a useful two-stage recovery problem. The total first determines a missing frequency; the width then determines the missing height. If you subtract the known heights from ninety, you mix counts with counts per unit width. The calculator will produce a number, but the expression has no coherent meaning. Write the units or the column labels before combining quantities.
Check the completed display by rebuilding the total: five times four, plus ten times three, plus twenty times two, equals ninety. A reverse check is especially valuable when a missing value was calculated through several relationships. It confirms that the inferred density fits both its class width and the total count given in the question.
There is also a limit to the method. If two bars have unknown frequencies and you know only their combined total, there may be many possible pairs of heights. For instance, a remaining frequency of forty could be split into ten and thirty or twenty and twenty. Without an additional relationship, the individual heights are not determined. Do not force a unique answer from one total when more than one unknown remains. State the missing condition, or show two compatible assignments to demonstrate why the information is insufficient.
These answers revisit the distinctions most likely to be confused during quick graph reading. After choosing the relevant answer, calculate one example using both the width and height. The aim is to attach the formula to a visible relationship, rather than remember three words without knowing which quantity each represents.
Is the tallest bar always the most frequent class?
Not if “most frequent” means the greatest class count and widths differ. In a frequency-density histogram, the tallest bar has the greatest density. Compare areas to find the greatest frequency. If widths are equal, height and frequency comparisons agree because every height is multiplied by the same width.
Can a frequency density be less than one?
Yes. Twelve observations across a width of twenty give density 0.6. Multiplying 0.6 by twenty returns twelve. A density is a rate across an interval, so it need not be an integer even when the original observations count people or other indivisible objects.
Does the flat bar mean observations are evenly spread?
No. The rectangle represents a grouped total through its area. It does not show the exact positions of the observations within the interval. An even spread can be an explicit assumption for estimation, but it is not proved by the horizontal top of the bar.
What should I check first in an unfamiliar histogram?
Read the variable, units, class boundaries and vertical-axis label. Then calculate one bar’s frequency to confirm the interpretation. Only after that compare counts, concentrations or proportions according to the question. A few seconds spent identifying the representation can prevent an entire solution based on the wrong feature.
Continue the Mathematics series
All G1/G2/G3 Mathematics explainers
What Is G1 Mathematics | Why Two Successive Discounts Cannot Simply Be Added
What Is G2 Mathematics | Why a Zero Product Does Not Mean a Zero Sum
What Is G1/G2/G3 Mathematics | Why a Rounding Upper Bound Is Usually Not an Included Value