4

enter image description here

The question asks to describe the distribution of aspen tree diameters from the sample.

I said that the distribution was bimodal with one peak around 5.2 and the other peak around 9.2.

However the correct answer is that the distribution is skewed to the right and has a gap between 7 and 8 inches.

I tried looking online for some answers but I can't find any. Why is this a skewed unimodal distribution instead of a bimodal distribution? Is it because in a bimodal distribution there are two peaks but there is no gap, instead it is just a very low frequency between the two peaks?

user130306
  • 1,890
  • You ask "Why is this a skewed unimodal distribution?", but nothing in the post suggests that it should be. – joriki May 24 '20 at 23:44
  • 1
    "Describe a distribution" is not a well defined activity. Given the same distribution, there are multiple ways of describing it. I think your answer is better than the "correct answer". I would have added that about $3/4$ of the trees were in the lower group and felt it was an excellent description. It gives me a much better image of the distribution than the correct answer. The question was probably written by somebody who thinks all distributions are normal, so you should describe how this one varies from that. – Ross Millikan May 25 '20 at 04:14

2 Answers2

1
  • Can a bimodal distribution has a gap.

Yes, it is possible to have $0$ probability in between.

For example, consider the following discrete random variable.

$$P(X=1)=0.1, P(X=2)=0.3, P(X=3)=0.1$$

$$P(X=8)=0.1, P(X=9)=0.3, P(X=10)=0.1$$

It is bimodal and it can't take value from $4$ to $7$.

Your distribution is skewed to the right as there are more points on the left.

Your distribution is skewed to the right, bimodal, and has a gap.

Siong Thye Goh
  • 149,520
  • 20
  • 88
  • 149
1

This question appears online in several places, and the section you're quoting is part b). In part a) it explains that the tree diameters in 1975 had a normal distribution. In that context, it may make sense to just describe this distribution as having a gap and being skewed right (relative to the previous distribution). Without that context, your answer seems absolutely correct... it looks bimodal.

That all being said, part e) of the same question notes that trees grow smaller in the highlands than in the lowlands of the park, and expresses concern that the highlands were oversampled... so above and beyond the appearance of the distribution, there's evidence that the distribution really is bimodal!

mjqxxxx
  • 41,358