I used to know this 25-cent word, and I have forgotten it. It describes a phenomenon where common descriptive statistics like mean and variance misrepresent the individuals or maybe the phenomenon itself. (I don't know because I can't look it up!) I believe it begins with an e, probably an en- or eu-. Does it sound familiar to anyone?
-
$\begingroup$ Could be any number of things (bias? insufficiency? inefficiency?). It might help if you gave more context, at present this is probably too broad (too many possible answers) $\endgroup$– Glen_bJul 3, 2018 at 2:38
-
$\begingroup$ Unrepresentative or non-representative perhaps? $\endgroup$– JimJul 3, 2018 at 9:38
-
$\begingroup$ Thanks for trying; it's more obscure than bias and the like. Encastic, euthentical, endostatic, something like that. "Children's classroom learning should not be reduced to test scores because children's individual educational experiences are . . . . eunomical . . . . or something." Looking for this one single specific word, not any word that describes this situation. $\endgroup$– J KellyJul 3, 2018 at 15:21
-
$\begingroup$ Could it be Exogenous? en.wikipedia.org/wiki/Exogeny $\endgroup$– RLaveJul 4, 2018 at 13:55
2 Answers
The word I'd been driving myself crazy to recall was "ergodic" -- "the ensemble average equals the time average" -- with the concept I was trying to research being "nonergodic" meaning a phenomenon dependent on its history.
Based on the sound of the word, maybe idiosyncratic? If there is idiosyncratic error, then estimating summary statistics just in a cross-section would not be meaningful.