What does "a proxy" mean in the context of this statistical joke?

Question

“Let us remember the unfortunate econometrician who, in one of the major functions of his system, had to use a proxy for risk and a dummy for sex.”

I understand why the second part is funny (a dummy for sex can be understood as a sex doll), but I do not get the first part. I found that a proxy can mean either

1. the authority to represent someone else, especially in voting; or
2. a figure that can be used to represent the value of something in a calculation

It was the second meaning that the econometrician had in mind. But how does the first meaning fit in with risk? Or is there another meaning that makes a proxy for risk funny?

Answer 1

The "proxy for risk" part is not really meant to be funny. It would be fairly routine to describe an econometrician as "unfortunate" because he was unable to measure risk directly, but had to rely on a proxy (that is, a numerical value that stands in as a substitute for a direct measurement) instead. This would make his thesis less secure because people could question whether the proxy was valid or not.

The joke is that "proxy for risk" is a reasonable misfortune for a econometrician, but "dummy for sex" is has no relation to econometrics. But the two statements have an exact parallel construction, along with the parallel meaning of using a substitute for the real thing: in the first case a numerical model instead of a direct measure; and in the second case a doll instead of a real person.

Jokes are often constructed by directly connecting a reasonable statement with a ridiculous one.

Answer 2

The econometrician was unfortunate because he had to use an observed variable (proxy) to represent an unobserved variable (risk). Risk is the probability/possibility of something bad happening. You cannot use something that actually happened to represent something that can possibly happen. On the second part, the funny part is on how he collected his/her data in such a way that sex would be coded as a dummy variable i.e. 1=sex and 0=non-event. If he/she used an experimental approach, that means he/she had to allow one group of his total sample to have sex and code 1 for all individuals who would have participated in the act of sex. The other group that would not be allowed to have sex would be treated as the control group. If the researcher collected data through questionnaires, that means he/she needed to have asked the direct question: Have you had sex? Yes/No. Then code yes=1 and no=0 so that the no group would constitute the baseline group of the dummy variable. You will be literally asking research participants if they removed their pants.