My girlfriend is an Actuarial Analyst at a large insurance company in the Netherlands and because we'll soon have our two year anniversary, I thought of gifts for her.

On Proof: Math is beautiful I discovered these Distribution pluffies.

So here's my question: What distribution is of the most relevance in the field of an econometrician?

The available pluffies are:

  • Standard Normal Distribution
  • t Distribution
  • Chi-Square Distribution
  • Log Normal Distribution
  • Continuous Uniform Distribution
  • Weibull Distribution
  • Cauchy Distribution
  • Poisson Distribution
  • Gumbel Distribution
  • Erlang Distribution

Any help much appreciated.

EDIT: Thanks a lot for all the suggestions despite this being just off-topic! I'll get her the t Distribution pluffy.

  • $\begingroup$ Buy her a vacuum; chicks love gifts like that. $\endgroup$ – Joshua Ulrich May 4 '11 at 19:19
  • $\begingroup$ Sounds silly, but how about choosing on the basis of colour? (Otherwise, I would probably second the Cauchy choice -- although a positive distribution with infinite mean might be better if you want to indicate something about the depth of your affection ...) $\endgroup$ – Ben Bolker May 4 '11 at 19:27

You gotta get her one with some Kurtosis. Maybe the t-distribution. And be sure and write a loving note along the lines of, "Baby, when I think of fat tails, I think of you. Your kurtosis makes you non-normal."

My wife digs it when I get sappy like that. I have the scars to prove it.

  • $\begingroup$ So, when is the divorce final? $\endgroup$ – bill_080 May 4 '11 at 23:58
  • $\begingroup$ No such thing as too many tail puns. $\endgroup$ – Matt Parker May 5 '11 at 1:36
  • $\begingroup$ Liked the suggestion about the note. Thanks for your help. $\endgroup$ – Philipp May 5 '11 at 8:26

You're in big trouble if you're asking us for gift advice.

  • $\begingroup$ True, we don't have a vested interest and we may want to be funny (which may appear we are mean). Unless you want to cut us in a deal. $\endgroup$ – Roman Luštrik May 4 '11 at 21:17

Insurance is all about skewed distributions with long tails: think amount of loss. These also typically have only positive values. The log-normal distribution looks most like one of those. Another good option is the Gumbel distribution, which comes up in extreme value theory.


Aren't econometricians concerned with the price of t (distributions) in China? It has the large (on occasion, infinite) kurtosis recommended by @JD Long, too.


From the list I would pick standard normal. After all regression is the main tool of econometrician and usually econometrician can rely only on asymptotic results, hence standard normal rules them all :)

Having said that I would not like to get a standard normal distribution pluffy (I am not a girl, but can be considered econometrician) for standard normal is so widely used that it is a bit mundane. So you can choose log-normal, since econometricians use log-log regression extensively or Cauchy as a reminder, that not all distributions have finite moments.

  • $\begingroup$ His GF could misunderstand Cauchy: "or Cauchy as a reminder, that not all distributions have finite moments.". :P $\endgroup$ – Roman Luštrik May 4 '11 at 18:52
  • $\begingroup$ @Roman, good catch :) $\endgroup$ – mpiktas May 4 '11 at 18:53

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