7
$\begingroup$

Is it possible to have a right-skewed distribution with mean equal to mode? If so, could you give me some example?

$\endgroup$
  • 3
    $\begingroup$ Take a suitable mixture of any finite-mean skewed distribution and any finite-mean unimodal symmetric distribution with the same mean. All continuous examples and all discrete examples arise in this way. $\endgroup$ – whuber Apr 15 at 20:27
  • $\begingroup$ @whuber, That's a great idea. If you have time, it would be terrific if you made a slightly more detailed answer out of that. $\endgroup$ – beta1_equals_beta2 Apr 15 at 21:25
10
$\begingroup$

Easy examples come from binomial distributions -- which can hardly be dismissed as pathological or as bizarre counter-examples constructed ad hoc. Here is one for 10 trials and probability of success 0.1. Then the mean is 10 $\times$ 0.1 = 1, and 1 also is the mode (and for a bonus the median too), but the distribution is manifestly right skewed.

The code giving the number of successes 0 to 10 and their probabilities 0.348678... and so forth is Mata code from Stata, but your favourite statistical platform should be able to do it. (If not, you need a new favourite.)

: (0::10), binomialp(10, (0::10), 0.1)
                  1             2
     +-----------------------------+
   1 |            0   .3486784401  |
   2 |            1    .387420489  |
   3 |            2   .1937102445  |
   4 |            3    .057395628  |
   5 |            4    .011160261  |
   6 |            5   .0014880348  |
   7 |            6    .000137781  |
   8 |            7   8.74800e-06  |
   9 |            8   3.64500e-07  |
  10 |            9   9.00000e-09  |
  11 |           10   1.00000e-10  |
     +-----------------------------+

Among continuous distributions, the Weibull distribution can show equal mean and mode yet be right-skewed.

$\endgroup$
  • $\begingroup$ Thank you! This is absolutely helpful! I will also look into the Weibull distribution you mentioned. $\endgroup$ – Don Tawanpitak Apr 15 at 19:00
  • $\begingroup$ By the way, do you know some other continuous distribution with finite support that can exhibit the same property? $\endgroup$ – Don Tawanpitak Apr 15 at 19:04
  • 2
    $\begingroup$ @DonTawanpitak A quick numerical search for the Weibull only revealed one solution: $\alpha = 3.3125, \beta = 1$ where $\alpha$ is the shape and $\beta$ is the scale. The mode and mean are then $0.897186$. But this Weibull isn't terribly right skewed (its skewness is $0.074$). $\endgroup$ – COOLSerdash Apr 15 at 19:12
4
$\begingroup$

If the distribution is discrete, sure. It's easy. For example, a distribution with probability mass function

  • $P(X=0) = 0.36$
  • $P(X=1) = 0.40$
  • $P(X=2) = 0.13$
  • $P(X=3) = 0.10$
  • $P(X=4) = 0.01$

is right (i.e. positively) skewed and has both a mean and a mode of 1.

$\endgroup$
  • $\begingroup$ Is this suggesting that discrete distributions, as opposed to continuous distributions are more likely to exhibit this property? There seems to be a strong argument for that. $\endgroup$ – Thomas Cleberg Apr 15 at 18:47
  • 1
    $\begingroup$ No, I'm not suggesting that. I'm just saying it's easy to come up with an example (which is all the OP asked for) of a discrete distribution with that property. $\endgroup$ – beta1_equals_beta2 Apr 15 at 18:53

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.