If $R$ and $s$ are the range and standard deviation of a set of $n$ values, then $4\le R^2/s^2\le 2n$

1.5k Views Asked by At

Problem

Let $R$ and $s$ be respectively the range and standard deviation , respectively, of a set of $n$ values. Then prove that $\dfrac{R^2}{2n}\leq s^2\leq \dfrac{R^2}{4}$.

I have tried using the definition of range and standard deviation but neither of the inequality can be proved. Can anyone help me in proving it?

1

There are 1 best solutions below

2
On BEST ANSWER

Without loss of generality (by translation), assume that the data is such that $0\leq x_i\leq R$, with the smallest data $x_{\min}=0$, and the largest data is $x_{\max}=R$. Let $\mu$ be the mean. Then $$\sum_i x_i^2 = \sum_i x_ix_i \leq \sum_i Rx_i= nR\mu$$ Also note that $\mu\leq R$. Therefore, $$ns^2= \sum_i (x_i-\mu)^2 = \sum_i x_i^2 -n\mu^2 \leq nR\mu-n\mu^2 =n\mu(R-\mu)\leq nR^2/4.$$ So, $s^2\leq R^2/4$.

For the lower bound, note that $$2ns^2= 2\sum_i (x_i-\mu)^2 \geq 2[(x_{\max}-\mu)^2+ (x_{\min}-\mu)^2]=2[(R-\mu)^2+\mu^2]\geq R^2,$$ where the last inequality follows from arithmetic-quadratic mean inequality, which is $$\frac{x_1^2+\ldots+x_n^2}{n} \geq \left(\frac{x_1+\ldots+x_n}{n}\right)^2$$ applied to $(R-\mu)$ and $\mu$. Therefore, $s^2\geq R^2/2n$, and so this completes the proof.