Ho do we prove that the standard deviation is greater than or equal to the mean deviation about the arithmetic mean ?

$$ \sqrt\frac{\sum_{i=1}^{n}(x_i-\bar{x})^2}{n}\geq\frac{\sum_{i=1}^{n}|x_i-\bar{x}|}{n} $$

and under what conditions we get the equality ?

I think i understand that it is because of the squaring in standard deviation which tends to give more weightage to the data far from the central tendency.

  • 3
    $\begingroup$ The Cauchy–Schwarz inequality will do. $\endgroup$ – Chappers Jul 20 '17 at 18:18
  • $\begingroup$ @Chappers could u help me how to proceed. i'm unable to find where to start inorder to prove it. $\endgroup$ – ss1729 Jul 20 '17 at 18:41

Let $v=\left|\vec{x} - \bar{x}\vec{1}\right|$, where $|\cdot|$ is component-wise. Then: \begin{align} \frac{1}{n^2}\left( \sum_i |x_i - \bar{x}| \right)^2 &= \frac{ \left(v\cdot\vec{1}\right)^2}{n^2}\\ &\leq \frac{(\vec{1}\cdot\vec{1})(v\cdot v)}{n^2}\\ &= \frac{1}{n}\sum_i|x_i - \bar{x}|^2 \end{align} where we used the CS inequality for the second step. Now take the root of the first and last terms: $$ \sqrt{\frac{1}{n}\sum_i|x_i - \bar{x}|^2\;} \,\geq \frac{1}{n}\sum_i |x_i - \bar{x}| $$

| cite | improve this answer | |
  • $\begingroup$ How does $\left|\vec{x} - \bar{x}\vec{1}\right|= \sum_i \left|x_i - \bar{x} \right|$ ? $\endgroup$ – Revolucion for Monica Jul 21 '17 at 8:22

The Cauchy Schwarz inequality

$$\bigg(\sum_{i=1}^{n}a_{i}^2\bigg).\bigg(\sum_{i=1}^{n}b_{i}^2\bigg)\ge \bigg(\sum_{i=1}^{n}a_ib_i\bigg)^2 $$

taking $a_i=|x_i-\bar{x}|$ and $b_i=1/n$,

$$ \bigg(\sum_{i=1}^{n}|x_i-\bar{x}|^2\bigg).\bigg(\sum_{i=1}^{n}\tfrac{1}{n^2}\bigg)\ge \bigg(\sum_{i=1}^{n}|x_i-\bar{x}|.\tfrac{1}{n}\bigg)^2\\ \bigg(\sum_{i=1}^{n}(x_i-\bar{x})^2\bigg).\bigg(n.\tfrac{1}{n^2}\bigg)\ge \bigg(\frac{\sum_{i=1}^{n}|x_i-\bar{x}|}{{n}}\bigg)^2\\ \frac{\sum_{i=1}^n(x_i-\bar{x})^2}{n}\ge \bigg(\frac{\sum_{i=1}^n|x_i-\bar{x}|}{n}\bigg)^2\\ \sqrt\frac{\sum_{i=1}^n(x_i-\bar{x})^2}{n}\ge\frac{\sum_{i=1}^n|x_i-\bar{x}|}{n}\\ S.D\ge M.D $$

| cite | improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.