Dear Karl,
Thanks for your reaction.
> I think one could argue that a "sample" taken *on* the second
> is most representative of an interval extending from half a
> second prior to the sample time and half a second following the
> sample time, [?]
>
> On the other hand, if there is some finite response response
> time of your instrument (say, of order 1 sec), you could argue
> that a sample taken at time t really represents an average over
> some preceding interval, [?]
I hadn't considered taking the sampling times as representatives of
lower-level intervals. That will often be the case. But I think that to
avoid having to know all details of the underlying data generation process,
we may assume for purposes of the bounds variable for the derived
statistics that the samples are instantaneous. (Feel free to correct me.)
Best,
Erik
--
https://ac.erikquaeghebeur.name
Received on Mon Mar 26 2018 - 15:21:36 BST
