7

Question basically in title. I am just curious if anyone has any historical references that point to the earliest usages of this notation.

nth
  • 766
  • 4
  • 9
  • See https://stats.stackexchange.com/questions/41306/why-are-probability-distributions-denoted-with-a-tilde/41332#41332 – kjetil b halvorsen Mar 19 '18 at 13:20
  • @kjetilbhalvorsen Sure, but that answer pertains to matrices and matrices have a history dating back 2000 years. Modern statistical notation was largely developed in the 20th century. – nth Mar 19 '18 at 13:46
  • Well, but that was not the point. The point was that conventions are conventions, they are what they are, good or bad, and asking Why does' nt make much sense. Asking about history is ifferent. – kjetil b halvorsen Mar 19 '18 at 13:55
  • There are some close votes. I do not understand why this is opinion-based, it is asking for historical references, which is clear and on-topic – kjetil b halvorsen Mar 19 '18 at 14:06
  • 3
    I'm tempted to guess n is short for number, but not sure. I do know that it's not upper case to distinguish it from population size (N). – MBorg Mar 19 '18 at 14:07
  • It is not opinion-based but it can considered off topic because it doesn't really pertain to issues in statistics, data mining or machine learning. Also although answers to this could be known if there is an historical reason, it seems that we are only guessing. – Michael R. Chernick Mar 19 '18 at 14:59
  • It is not in David's lists of first occurrences of statistical terms (either 1995 one or 1998). However since American Statistician was willing to publish those lists this question does seem to be to be on topic. – mdewey Mar 19 '18 at 15:43
  • 4
    The notation comes directly from mathematics -- collections of things in mathematics are frequently counted $1,2,...,n$. So, for example, you can see Pearson in 1900 (in ordinary use of mathematical notation) referring to a collection of deviations from expected as $x_1, x_2, ..., x_n$ (in a goodness of fit situation). From that standard use of mathematical notation to labelling original observations in a similar fashion, and so having the sample size be $n$ is an obvious step. I'd say you would have to go back to the first use of $n$ in mathematics to refer to an unspecified number of objects – Glen_b Mar 19 '18 at 21:40
  • ... so for example, you have Cayley in 1843 referring to $n$ dimensions, and this is the same sense in which Pearson was using $n$ in 1900. – Glen_b Mar 19 '18 at 21:44

0 Answers0