RankRizz
JEE MAIN 2020 · 03 SEP S1+4 / −1HARDTRANSCRIBED

JEE Main 2020 Mathematics question — Variance and standard deviation

For the frequency distribution: Variate (x)(x): x1,x2,x3,…,x15x_1, x_2, x_3, \ldots, x_{15}
Frequency (f)(f): f1,f2,f3,…,f15f_1, f_2, f_3, \ldots, f_{15}
where 0<x1<x2<x3<…<x15=100 < x_1 < x_2 < x_3 < \ldots < x_{15} = 10 and ∑i=115fi>0\sum_{i=1}^{15} f_i > 0, the standard deviation cannot be

SOLUTION

We have a grouped (but discrete) frequency distribution whose observed values satisfy 0<x1<x2<⋯<x15=10.0<x_1<x_2<\dots<x_{15}=10. Thus every variate value lies strictly between a positive number (call it m=x1m=x_1) and 1010. Hence each xix_i is contained in the closed interval [ m, 10 ][\,m,\,10\,] where 0<m<10.0<m<10.

Let the total frequency be N=∑i=115fi,N=\displaystyle\sum_{i=1}^{15}f_i, which is given to be positive. The (arithmetic) mean of the distribution is

xˉ  =  ∑i=115fi xi∑i=115fi  =  ∑fi xiN.\bar x \;=\;\dfrac{\sum_{i=1}^{15}f_i\,x_i}{\sum_{i=1}^{15}f_i} \;=\;\dfrac{\sum f_i\,x_i}{N}.

The population standard deviation (about the mean) is defined by the well-known formula

σ  =  ∑i=115fi (xi−xˉ)2∑i=115fi  =  ∑fi (xi−xˉ)2N.\sigma \;=\;\sqrt{\dfrac{\sum_{i=1}^{15}f_i\,(x_i-\bar x)^2}{\sum_{i=1}^{15}f_i}} \;=\;\sqrt{\dfrac{\sum f_i\,(x_i-\bar x)^2}{N}}.

To find an upper bound on σ\sigma we invoke an elementary but very useful inequality for any set of numbers lying in a fixed interval. Let the interval be [a,b][a,b] and denote the mean by μ.\mu. Since every observation satisfies a≤x≤b,a\le x\le b, we can write

x−μ  ≤  b−μandμ−x  ≤  μ−a.x-\mu \;\le\; b-\mu\quad\text{and}\quad \mu-x \;\le\; \mu-a.

Consequently the squared deviation satisfies

0  ≤  (x−μ)2  ≤  max⁡{(b−μ)2, (μ−a)2}.0\;\le\;(x-\mu)^2\;\le\;\max\{(b-\mu)^2,\,(\mu-a)^2\}. Because μ\mu itself is inside the interval [a,b],[a,b], both distances b−μb-\mu and μ−a\mu-a are at most the total length b−ab-a. Furthermore, at least one of them is not larger than half that length; precisely, we always have

max⁡{ b−μ, μ−a }  ≤  b−a2.\max\{\,b-\mu,\,\mu-a\,\}\;\le\;\dfrac{b-a}{2}.

Squaring this inequality gives

(x−μ)2  ≤  (b−a2)2.(x-\mu)^2\;\le\;\left(\dfrac{b-a}{2}\right)^2.

Averaging the above inequality over all observations can never increase the right-hand side; therefore

σ2=∑fi (xi−μ)2N  ≤  (b−a2)2,\sigma^2 =\dfrac{\sum f_i\,(x_i-\mu)^2}{N} \;\le\;\left(\dfrac{b-a}{2}\right)^2, or equivalently

σ  ≤  b−a2.\sigma\;\le\;\dfrac{b-a}{2}.

This result—which is sometimes quoted as “the variance of any sample contained in an interval of length LL cannot exceed L24\dfrac{L^2}{4}”—tells us that the standard deviation never exceeds half the range.

Now we apply the inequality to the present question. Our lower bound is a=ma=m and our upper bound is b=10.b=10. Therefore

σ  ≤   10−m 2.\sigma\;\le\;\dfrac{\,10-m\,}{2}.

Because m>0,m>0, we have 10−m<10,10-m<10, and hence the right-hand side is strictly less than 102=5.\dfrac{10}{2}=5. Putting it plainly,

σ  <  5.\sigma\;<\;5.

Thus no matter how the actual frequencies f1,f2,…,f15f_1,f_2,\dots,f_{15} are chosen (provided they are all positive), the standard deviation of the distribution can never reach 5,5, let alone exceed it.

Let us compare this strict upper bound with the four numerical options offered in the question:

Option A: 4(possible, because 4<5)4 \quad(\text{possible, because }4<5)
Option B: 1(certainly possible)1 \quad(\text{certainly possible})
Option C: 6(impossible, because 6>5)6 \quad(\text{impossible, because }6>5)
Option D: 2(possible)2 \quad(\text{possible})

Only the value 66 violates the theoretical ceiling of 55, so it is the unique value that the standard deviation cannot take.

Hence, the correct answer is Option C.

Tagged Variance and standard deviation. Machine-assigned, so read it as a strong shortlist rather than a verdict.

ChapterStatistics
TopicVariance and standard deviation
TypeMultiple choice