Free Mathematics Class 10 CBSE notes · practise this chapter with an AI quiz

← All study notes

The Middle Value and the Most Common Value Can Sit in Different Places

Build a cumulative frequency table, locate the median class and compute the median with its formula, identify the modal class and compute the mode, recover a missing frequency from a given median, and decide which average describes a real data set best.

Why does one data set need three different kinds of average?

Suppose a small workshop pays nine workers ₹1200012000 a month each and one owner ₹300000300000. The mean salary is ₹4080040800 — a figure that describes nobody in the building. Nine of the ten people earn less than a third of it.

The median, the middle value when the salaries are lined up, is ₹1200012000. So is the mode, the most frequent value. Those two describe the workshop honestly and the mean does not, and the reason is that the mean is dragged upward by one extreme value while the other two are not.

So the three averages answer three different questions:

-
The mean — if the total were shared equally, how much would each get?
-
The median — what is the value in the middle, with half the data on either side?
-
The mode — which value occurs most often?

None of them is the right answer in general. The mean uses every observation and is best when the data has no extreme outliers. The median ignores how far out the extremes are and so survives them. The mode is the only one that makes sense for data that is not numerical at all, such as the most common shoe size sold.

Part 1 dealt with the mean. This part gives you the other two for grouped data, and both need a new idea first: the cumulative frequency, which tells you how many observations fall below a given point. You cannot find a median without it, because the median is defined by a position rather than by a value.

And then one relation ties all three together approximately:


3 Median≈Mode+2 Mean3\,\text{Median} \approx \text{Mode} + 2\,\text{Mean}

which is useful as a rough check and as a way of estimating the third when two are known.

This page covers the second part of the CBSE Class 10 Maths chapter on statistics: cumulative frequency, the median and mode of grouped data, missing frequencies, and comparing the three averages.

How do you build a cumulative frequency table?

Add each frequency to the running total of all the frequencies before it. The result tells you how many observations are less than the upper limit of each class.

Worked example — the distribution used throughout this page. A test was taken by 5555 students and their marks recorded as follows.

- 00–1010: frequency 55
- 1010–2020: frequency 88
- 2020–3030: frequency 2020
- 3030–4040: frequency 1515
- 4040–5050: frequency 77

The less-than cumulative frequencies, each obtained by adding the next frequency to the previous total:

- Less than 1010: 55
- Less than 2020: 5+8=135 + 8 = 13
- Less than 3030: 13+20=3313 + 20 = 33
- Less than 4040: 33+15=4833 + 15 = 48
- Less than 5050: 48+7=5548 + 7 = 55

The final cumulative frequency must equal the total number of observations, and here it does: 5555. That is the compulsory check on any cumulative column — if the last entry is not nn, a frequency has been added wrongly or skipped.

Reading the table. "Less than 3030 is 3333" means 3333 of the 5555 students scored under 3030 marks. It says nothing about how those 3333 are distributed inside the classes below 3030; it only counts them.

There is also a more-than type, built by adding from the bottom up:

- More than or equal to 00: 5555
- More than or equal to 1010: 5050
- More than or equal to 2020: 4242
- More than or equal to 3030: 2222
- More than or equal to 4040: 77

Check that the two types agree. At any boundary the two must add to nn: less than 3030 is 3333, more than or equal to 3030 is 2222, and 33+22=5533 + 22 = 55. Correct.

Why the median needs this table and the mean did not. The mean was a total divided by a count, so the order of the data was irrelevant. The median is defined by position — the value with half the data below it — so you must be able to say how many observations lie below any point. That is exactly what a cumulative frequency is.

One boundary case about the two types. A question may give you a cumulative table and ask you to recover the ordinary frequencies. Subtract each entry from the next one: from 55, 1313, 3333, 4848, 5555 you get 55, 88, 2020, 1515, 77 — the original frequencies. Converting in both directions is examinable, and the give-away that a table is cumulative is that its entries never decrease.
Formula

What is the median formula for grouped data and how do you use it?

Find the class in which the middle observation falls, then interpolate inside it.

Median=l+(n2−cff)×h\boxed{\text{Median} = l + \left(\frac{\frac{n}{2} - cf}{f}\right) \times h}

where

- ll is the lower limit of the median class
-
nn is the total number of observations
-
cfcf is the cumulative frequency of the class immediately before the median class
-
ff is the frequency of the median class itself
-
hh is the class width

How to find the median class.
Compute n2\frac{n}{2} and look down the cumulative frequency column for the first entry that is greater than or equal to it. That class is the median class.

Worked example. Find the median of the distribution of 5555 marks tabulated above.

Step 1 — the position.

n2=552=27.5\frac{n}{2} = \frac{55}{2} = 27.5

Step 2 — the median class. The cumulative frequencies are 55, 1313, 3333, 4848, 5555. The first one to reach 27.527.5 is 3333, belonging to the class 2020–3030.

Step 3 — read off the four quantities.


- l=20l = 20, the lower limit of 2020–3030
- cf=13cf = 13, the cumulative frequency of the previous class 1010–2020
- f=20f = 20, the frequency of 2020–3030 itself
- h=10h = 10, the class width

Step 4 — substitute.

Median=20+(27.5−1320)×10=20+14.520×10=20+7.25=27.25\text{Median} = 20 + \left(\frac{27.5 - 13}{20}\right) \times 10 = 20 + \frac{14.5}{20} \times 10 = 20 + 7.25 = 27.25

Check that the answer lies inside the median class. It must, by construction, and 27.2527.25 does lie between 2020 and 3030. An answer outside the median class means one of the four quantities was misread — most often cfcf, by taking the median class's own cumulative frequency instead of the previous one.

What the formula is actually doing. There are 1313 students below 2020 marks and we want the 27.527.5th, so we need 14.514.5 more students. The class 2020–3030 contains 2020 students spread across 1010 marks, so 14.514.5 students' worth of it is 14.520\frac{14.5}{20} of the way across, which is 7.257.25 marks. Starting at 2020 and moving 7.257.25 gives 27.2527.25. The formula is nothing more than that sentence written in symbols, and reconstructing it that way is safer than remembering it.

Two quantities that are easy to confuse.

-
cfcf is the cumulative frequency of the class before the median class
, not of the median class. Here it is 1313, not 3333
- ff is the plain frequency of the median class, not a cumulative one. Here it is 2020, not 3333

Using 3333 for cfcf would give 20+27.5−3320×10=20−2.75=17.2520 + \frac{27.5 - 33}{20} \times 10 = 20 - 2.75 = 17.25, which falls outside the median class — and that is exactly why the plausibility check catches this particular error every time.

One note on n2\frac{n}{2}. For grouped data the position used is n2\frac{n}{2} whether nn is odd or even. There is no averaging of two middle values as there is for ungrouped data, because the data is continuous and the formula interpolates rather than picking an observation.

How do you find the mode of grouped data?

Find the class with the highest frequency, then interpolate inside it using how much taller it is than its two neighbours.

Mode=l+(f1−f02f1−f0−f2)×h\text{Mode} = l + \left(\frac{f_1 - f_0}{2f_1 - f_0 - f_2}\right) \times h

where

- ll is the lower limit of the modal class
-
f1f_1 is the frequency of the modal class
-
f0f_0 is the frequency of the class before it
-
f2f_2 is the frequency of the class after it
-
hh is the class width

Worked example.
Find the mode of the same distribution of 5555 marks.

Step 1 — the modal class. The frequencies are 55, 88, 2020, 1515, 77, and the largest is 2020. The modal class is 2020–3030.

Step 2 — read off the quantities.
l=20l = 20, f1=20f_1 = 20, f0=8f_0 = 8, f2=15f_2 = 15, h=10h = 10.

Step 3 — substitute.

Mode=20+(20−82(20)−8−15)×10=20+(1240−23)×10\text{Mode} = 20 + \left(\frac{20 - 8}{2(20) - 8 - 15}\right) \times 10 = 20 + \left(\frac{12}{40 - 23}\right) \times 10
=20+1217×10=20+7.06=27.06= 20 + \frac{12}{17} \times 10 = 20 + 7.06 = 27.06

Check that the mode lies inside the modal class: 27.0627.06 is between 2020 and 3030. Correct.

What the formula is doing. The modal class is taller than the class before it by 1212 and taller than the class after it by 55. The mode is pulled toward the taller neighbour, and the fraction 1212+5=1217\frac{12}{12 + 5} = \frac{12}{17} is the share of the class width you move across. Written that way the denominator makes sense, because

2f1−f0−f2=(f1−f0)+(f1−f2)=12+5=172f_1 - f_0 - f_2 = (f_1 - f_0) + (f_1 - f_2) = 12 + 5 = 17

So the denominator is the sum of the two differences, and computing it that way is much less error-prone than computing 2f1−f0−f22f_1 - f_0 - f_2 directly.

A useful consequence to sanity-check with. If the two neighbours have equal frequencies, the two differences are equal, the fraction is 12\frac{1}{2}, and the mode sits exactly at the middle of the modal class. The mode leans toward whichever neighbour is taller, and if it comes out on the wrong side, f0f_0 and f2f_2 have been swapped.

Worked example 2 — the modal class at an end. Suppose the frequencies were 2020, 1515, 88, 55, 77 instead, so the modal class is the first one, 00–1010.

There is no class before it, so f0=0f_0 = 0:

Mode=0+(20−02(20)−0−15)×10=2025×10=8\text{Mode} = 0 + \left(\frac{20 - 0}{2(20) - 0 - 15}\right) \times 10 = \frac{20}{25} \times 10 = 8

Taking the missing neighbour's frequency as zero is the convention, and it is what a question with a modal first or last class expects.

When the mode is not well defined. If two classes share the highest frequency, the distribution has two modal classes and the formula cannot be applied to either without ambiguity. Such data is called bimodal, and a question will not ask for a single mode from it — but recognising the situation is worth a mark if asked.

Putting all three averages together for this data. The mean of the same 5555 marks, by the direct method with class marks 55, 1515, 2525, 3535, 4545:

∑fixi=25+120+500+525+315=1485,xˉ=148555=27\sum f_i x_i = 25 + 120 + 500 + 525 + 315 = 1485, \qquad \bar{x} = \frac{1485}{55} = 27

So the three averages are: mean 2727, median 27.2527.25, mode 27.0627.06 — all close together, which tells you the distribution is fairly symmetric and has no extreme outliers. Compare that with the workshop salaries of the opening section, where the mean was ₹4080040800 and the median ₹1200012000: a large gap between the mean and the median is the signature of a skewed data set.

And the empirical relation, checked on this data.

3 Median=3×27.25=81.753\,\text{Median} = 3 \times 27.25 = 81.75
Mode+2 Mean=27.06+54=81.06\text{Mode} + 2\,\text{Mean} = 27.06 + 54 = 81.06

Close but not exactly equal, which is why the relation is written with an approximation sign. It is a rule of thumb for estimating a third average from two, not an identity, and quoting it as an exact equation is a marked error.

How do you find missing frequencies when the median is given?

Use the total frequency to get one equation, and the median formula to get another. Two unknowns need two conditions, exactly as in the mean chapter.

Worked example. The median of the following distribution is 525525 and the total frequency is 100100. Find xx and yy.

- 00–100100: 22
- 100100–200200: 55
- 200200–300300: xx
- 300300–400400: 1212
- 400400–500500: 1717
- 500500–600600: 2020
- 600600–700700: yy
- 700700–800800: 99
- 800800–900900: 77
- 900900–10001000: 44

Step 1 — the total frequency gives the first equation.

2+5+x+12+17+20+y+9+7+4=1002 + 5 + x + 12 + 17 + 20 + y + 9 + 7 + 4 = 100
76+x+y=100⟹x+y=2476 + x + y = 100 \quad\Longrightarrow\quad x + y = 24

Step 2 — locate the median class from the given median. The median is 525525, which lies in the class 500500–600600. So l=500l = 500, f=20f = 20, h=100h = 100, and n2=50\frac{n}{2} = 50.

Step 3 — the cumulative frequency before the median class, in terms of xx:

cf=2+5+x+12+17=36+xcf = 2 + 5 + x + 12 + 17 = 36 + x

Step 4 — substitute into the median formula.

525=500+(50−(36+x)20)×100525 = 500 + \left(\frac{50 - (36 + x)}{20}\right) \times 100
25=(14−x20)×100=5(14−x)25 = \left(\frac{14 - x}{20}\right) \times 100 = 5(14 - x)
5=14−x⟹x=95 = 14 - x \quad\Longrightarrow\quad x = 9

Step 5 — the second unknown, from the first equation:

y=24−9=15y = 24 - 9 = 15

Check by recomputing the median. With x=9x = 9 the cumulative frequency before the median class is 36+9=4536 + 9 = 45, so

Median=500+(50−4520)×100=500+520×100=500+25=525\text{Median} = 500 + \left(\frac{50 - 45}{20}\right) \times 100 = 500 + \frac{5}{20} \times 100 = 500 + 25 = 525

Exactly the given median, and the frequencies now total 100100. Both conditions are satisfied.

Also confirm that 500500–600600 really is the median class with these values. The cumulative frequencies become 22, 77, 1616, 2828, 4545, 6565, 8080, 8989, 9696, 100100. The first to reach 5050 is 6565, belonging to 500500–600600. Consistent — and that check matters, because a value of xx that shifted the median class elsewhere would have invalidated the whole calculation.

The structure of every question in this family.

- One unknown frequency — the median or the mean alone is enough
- Two unknown frequencies — you need the total frequency as well, and the total gives the easier equation, so use it first
- Always identify the median class from the given median, not by guessing, since the median lies inside its own class by construction

Worked example 2 — one unknown, from the mode instead. Suppose the frequencies of 00–1010, 1010–2020, 2020–3030, 3030–4040 are 55, 88, ff, 1515, the modal class is 2020–3030, and the mode is 27.0627.06. Find ff.

With l=20l = 20, f0=8f_0 = 8, f2=15f_2 = 15 and h=10h = 10:

27.06=20+(f−82f−23)×1027.06 = 20 + \left(\frac{f - 8}{2f - 23}\right) \times 10
7.06=10(f−8)2f−237.06 = \frac{10(f - 8)}{2f - 23}

Trying f=20f = 20 gives 10×1217=7.06\frac{10 \times 12}{17} = 7.06. So f=20f = 20, which is the distribution of the previous section — as it should be, since that is where the mode 27.0627.06 came from.

Note that a mode equation is not linear in ff, because ff appears in the denominator too. That is why questions prefer the median or the mean for missing-frequency problems, and why a mode version usually supplies enough information to check a candidate value rather than to solve from scratch.
Exam tip

What does a full-mark median or mode answer contain?

Show the cumulative frequency column, name the median or modal class explicitly, list the four quantities with their values, and then substitute. Each of those is a separate mark.

- Build the cumulative column and check the last entry equals nn
- Compute n2\frac{n}{2} and state it before looking for the median class
-
Name the median class
: "the first cumulative frequency to reach 27.527.5 is 3333, so the median class is 2020–3030"
- Write ll, cfcf, ff and hh as a list, and take cfcf from the class before the median class
- For the mode, name the modal class as the one with the highest frequency, and list ll, f1f_1, f0f_0, f2f_2 and hh
- Compute the mode's denominator as (f1−f0)+(f1−f2)(f_1 - f_0) + (f_1 - f_2) rather than as 2f1−f0−f22f_1 - f_0 - f_2
- Take a missing neighbour's frequency as zero when the modal class is first or last
- Check the answer lies inside its own class — the median inside the median class, the mode inside the modal class
- Use the total frequency first when two frequencies are unknown
- Re-verify a found frequency by recomputing the median and confirming the median class has not moved
- Quote the empirical relation with an approximation sign, never as an equality

The misconception to name. The median class and the modal class are not always the same class, and when they coincide it is a property of the data rather than a rule. The median class is found from the cumulative frequencies; the modal class from the plain frequencies — and a distribution with a long tail can easily have them in different places. Using one class for both calculations without checking is a real error even when it happens to give the right answer here.

A second trap. Taking cfcf as the median class's own cumulative frequency. It must be the previous class's, and the error is self-announcing: it makes n2−cf\frac{n}{2} - cf negative, so the median comes out below the lower limit of its own class. In this chapter's data it gives 17.2517.25 for a median class of 2020–3030, which is impossible.
Did you know

Why does one very large salary move the mean but not the median?

Go back to the workshop: nine workers on ₹1200012000 and one owner on ₹300000300000. Now suppose the owner's pay doubles to ₹600000600000.

The mean jumps
from ₹4080040800 to ₹7080070800, because the total has risen by ₹300000300000 and been shared among ten people.

The median does not move at all. It is still ₹1200012000, because the middle of the list is still a worker's salary — and it would stay ₹1200012000 if the owner earned ten crore.

That difference has a clean explanation. The mean uses the value of every observation, so a value far from the rest pulls it. The median uses only the order, so moving the largest observation further out changes nothing except how far out it is. The median is said to be resistant to outliers, and the mean is not.

Which is why the choice of average is a real decision and not a matter of taste.

- To describe a typical income, rent or house price, the median is used, because a few very large values would otherwise distort the picture
- To compute a total from an average, the mean is the only one that works — the mean times the count gives the total, and the median does not
- To decide which size of shirt to stock most of, the mode is the only sensible answer, since you cannot stock a mean size

The mode has one advantage the others lack entirely. It applies to data that is not numbers at all. The most common blood group in a sample, the most common mode of travel to school, the most frequently sold flavour — none of these has a mean or a median, because you cannot add or order them meaningfully. The mode is defined for any kind of data.

And a small warning about the mode of grouped data. The formula returns a value inside the modal class, but the modal class itself depends on how you chose the classes. Regroup the same observations into classes of width 2020 instead of 1010 and the modal class may change, and the mode with it. The mean and the median are far less sensitive to the grouping — which is a genuine limitation of the mode and a good answer to "which average is least reliable for grouped data?"

One last observation about the empirical relation. 3 Median≈Mode+2 Mean3\,\text{Median} \approx \text{Mode} + 2\,\text{Mean} holds well for moderately skewed data and poorly for badly skewed data. Check it on the workshop: the mean is ₹4080040800, the median and mode are both ₹1200012000, so the left side is ₹3600036000 and the right side is ₹9360093600. Nowhere near. The relation is a rule of thumb whose failure is itself informative — when it fails badly, the data has extreme outliers, which is exactly the case where the mean should not have been used to describe the data in the first place.
Exam relevance

How are median and mode used in JEE and NEET?

This is foundation work for Class 11 Statistics, a JEE Main topic, and the data-description ideas here are used directly in NEET Biology and in laboratory work.

Where the median leads. Class 11 uses it to define the mean deviation about the median, and proves the useful fact that the mean deviation is least when taken about the median rather than about any other point. The positional definition you learn here is what makes that true, and JEE Main asks for mean deviation about the median as a direct numerical.

Where the cumulative frequency leads. It becomes the basis of quartiles, deciles and percentiles, all located by the same interpolation formula with n4\frac{n}{4} or kn100\frac{kn}{100} in place of n2\frac{n}{2}. The formula does not change at all — only the fraction of nn you look for in the cumulative column. Recognising that means the median formula is really one formula for a whole family.

Where the comparison of the three averages leads. Class 11 introduces the coefficient of variation to compare the consistency of two data sets, and the discussion of skewness begins exactly where this chapter ends: a gap between the mean and the median signals skewness. NEET questions on interpreting biological data rely on being able to say which average describes a sample fairly.

Where the outlier-resistance idea leads. It matters in every practical measurement. A single bad reading shifts the mean of a set of observations but not their median, which is why experimental work often reports both, and why a discordant reading must be identified rather than silently averaged in. That reasoning appears in JEE and NEET Physics error-analysis questions.

Where the mode leads. It is the average used for categorical data, which is the only kind available in much of Biology — blood groups, phenotypes, the most frequent allele. NEET Biology uses modal reasoning without naming it, and in genetics the most probable phenotype of a cross is a modal statement.

Question types to expect. At this level: cumulative frequency tables, median and mode calculations, missing frequencies from a given median, and choosing the appropriate average. In competitive papers: mean deviation about the mean and the median, variance and standard deviation, coefficient of variation, and interpretation questions.

The single trap that costs marks. Using the median class's own cumulative frequency as cfcf. It must be the previous class's, and the error shows up as a median lying below the median class — 17.2517.25 for a class of 2020–3030 in this chapter's data. A plausibility check catches it in one second.

A second trap. Quoting 3 Median=Mode+2 Mean3\,\text{Median} = \text{Mode} + 2\,\text{Mean} as an exact identity. It is approximate, holding well only for moderately skewed data — on this chapter's distribution it gives 81.7581.75 against 81.0681.06, and on the workshop salaries it fails completely. Write it with an approximation sign and say what it is for.

Board versus competitive emphasis. The CBSE paper marks the cumulative column, the named class, the listed quantities, the substitution and the conclusion; a competitive paper marks a single number, often a quartile or a standard deviation. The transferable habit is locating a value by its position in the cumulative column — because that one technique gives you the median, every quartile and every percentile, and it is the whole of the interpolation method you will use next year.
Key takeaways

What must you be able to do from this part?

One cumulative column, two formulas and one approximate relation.

- The mean shares the total equally, the median is the middle value, the mode is the most frequent
- The median resists outliers; the mean does not. A single very large value moves the mean and leaves the median where it was
- Cumulative frequency is the running total, and the last entry must equal nn
- Less-than and more-than entries at the same boundary add to nn, so 33+22=5533 + 22 = 55
- Recover plain frequencies from a cumulative table by subtracting consecutive entries
- Median =l+(n2−cff)×h= l + \left(\frac{\frac{n}{2} - cf}{f}\right) \times h, with cfcf from the class before the median class
- Find the median class as the first class whose cumulative frequency reaches n2\frac{n}{2}
- For the 5555-mark distribution: n2=27.5\frac{n}{2} = 27.5, median class 2020–3030, l=20l = 20, cf=13cf = 13, f=20f = 20, h=10h = 10, giving a median of 27.2527.25
- Using cf=33cf = 33 would give 17.2517.25, outside the median class — which is how that error announces itself
-
Mode
=l+(f1−f02f1−f0−f2)×h= l + \left(\frac{f_1 - f_0}{2f_1 - f_0 - f_2}\right) \times h, and the denominator is (f1−f0)+(f1−f2)(f_1 - f_0) + (f_1 - f_2)
- Same distribution: modal class 2020–3030, f1=20f_1 = 20, f0=8f_0 = 8, f2=15f_2 = 15, giving 20+1217×10=27.0620 + \frac{12}{17} \times 10 = 27.06
- Take a missing neighbour as zero when the modal class is first or last
- Equal neighbours put the mode at the middle of the modal class; otherwise it leans toward the taller one
- The mean of the same data is 148555=27\frac{1485}{55} = 27, so all three averages are close and the distribution is nearly symmetric
-
A large gap between the mean and the median signals a skewed data set
-
3 Median≈Mode+2 Mean3\,\text{Median} \approx \text{Mode} + 2\,\text{Mean}
— here 81.7581.75 against 81.0681.06, so approximate and not exact
- Two missing frequencies need two conditions: a median of 525525 with a total of 100100 gives x=9x = 9 and y=15y = 15
- Re-verify that the median class has not moved after substituting a found frequency
- The median class and the modal class need not be the same class
- The mode is the only average defined for non-numerical data, and the only one sensitive to how the classes were chosen

The sharpest self-test is one table, three averages. Take the 5555-mark distribution, compute the mean, the median and the mode yourself, and then test the empirical relation — and decide, from how close the three come out, whether you would be comfortable describing this class's performance by the mean alone.

Ready to put this into practice?

Create a personalized quiz on this exact topic — free to start.

Create your own quiz on Statistics — Part 2Create a free account
← Back to all articles