Re: The survival time of chocolates on hospital wards: covert observational study
But I'm not so sure that it's right to talk about an "initial rapid “grab” phase".
Insofar as the survival function is exponential, the distribution of survival times follows an exponential probability density function. That is characteristic of a Markov process in which the probability of an individual chocolate being eaten in the next second (say) is independent of how long it has already survived. In other words the process is entirely random.
The reason that the survival curve falls rapidly at first is is merely a consequence of the fact that random (exponentially-distributed) intervals have more short (below the mean) intervals than long intervals (though the latter occupy more of the time).
In fact the curves may not be exactly exponential. Could that perhaps be because many people prefer the red ones to the green triangles? I look forward to the follow-up study.
Colquhoun, D. Lectures on Biostatistics, Clarendon Press 1970 (available at http://www.dcscience.net/Lectures_on_biostatistics-ocr4.pdf )
Competing interests: No competing interests