I'm building a units based type for C++ which describes probabilities (using boost::units library.) While I know that the values of probabilities are dimensionless, I think it makes some sense in the same way that angles (another dimensionless quantity) are often given units. My question is given that percent is a reasonable unit for parts per 100, is there some such concept for parts per unit/one? Following the etymology of percent (from per centum) I arrived at per unus (latin for one) => perun. However, I would prefer not making one up.

  • $\begingroup$ Please give an example sentence where you would want to use this term. $\endgroup$
    – amoeba
    Commented Jun 2, 2016 at 13:11
  • $\begingroup$ proportion? fraction? probability? $\endgroup$
    – Glen_b
    Commented Jun 2, 2016 at 15:24

3 Answers 3

  • Proportion or fraction (if out of one) or percentage (if out of 100) seem pretty innocuous and easy to interpret. Other bases may work too, like parts-per-million, depending on your intended domain. For example, some fields report outcomes as X per 100,000 patients or parts per million.

  • You could label them as relative frequencies (or something like r_freq), though this (obviously) has an frequentist slant and might be a little bit odd if the probability is being used to describe something like a degree of belief. On the other hand, if you're explicitly encoding subjective beliefs, you could call them credences (or maybe cred for short).

  • There isn't anything weird about calling the units probability, in the same way we often label an angle as being "in radians" even though angles are, in fact, dimensionless.

  • You could coin your own term, though I have a strong but viseral dislike of "perunus"--it just looks weird. It's emtymologically tempting, but I would strongly suggest not overloading probit either--it's confusing.

  • In some situations, it might be more natural to use odds instead of probability, which you could label as odds or odds ratio. I suppose there's also the Hartley or ban/deciban too, though these aren't quite probabilities (and have a very old-fashioned, Bletchley-park sort of feeling to them).

Finally, make sure that you're distinguishing between probabilities (which are dimensionless) and probability densities, which have units of $\frac{1}{\text{whatever}}$ (and, obviously, may be larger than one).


I usually use the term "proportion".

  • 4
    $\begingroup$ Fraction also works $\endgroup$
    – wabbit
    Commented Jun 2, 2016 at 13:30

A percent is just another way to represent any fraction by scaling it to be out of 100. You could do this for your scenario since something like .25/1 is obviously 25% of 1. It seems like there isn't a special term since there doesn't need to be.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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