0

I would like to check my knowledge of foreign words by sampling a N words dictionnary: after checking randomly n words I would find that I know k of them (and do not know n-k).

How should I choose n so that the sampling is representative (= so that I could say I know N k/n words)?

WoJ
  • 836
  • @Twink: I do not really understand your comment. What is there to try? I am wondering for which n the sampling is representative enough ("representative enough" being probably some kind of consensus) to say that N k/n can be seen as the number of words I know. If N=20000 and n=2 the result is not really relevant. If n=20000 then I have an exact result. I am looking for the right n in-between (again, "right" being the accepted consensus) – WoJ Sep 25 '13 at 08:33

1 Answers1

0

There is not good answer for that question because everything depend on how many words you really know. Lets assume that you know all words from the dictionary then any n gives you right answer as n=k, but if you know only one word then you will have to check entire dictionary to be sure that you have right results. As you do not know how many words you know as you want to discover it then checking entire dictionary is probably the only way

awg005
  • 1
  • The point was to find this number of words I know. If I know one word only, n=1 will give me 0 or N expected-to-be-known words - a very inaccurate result. The larger n is the closer I come to know my actual knowledge. With n=N I have an exact result (which would be k=1, with an expected number of words I know equal to one (correct). – WoJ Sep 26 '13 at 10:33