The simple reasoning based on mean expectations
leads to correct results only when all probabilistic effects
are negligible, an approximation which holds,
generally speaking, only for `large numbers'.
Under this approximation
the numbers of individuals tagged as Positive
or Negative can be considered to
follow in a deterministic way from the assumptions,
one of which is the proportion of infectees. This number can
then be obtained inverting the deterministic relation, thus yielding
Eq. (). But when fluctuations around the mean expectations
become important we need to use probability theory
in order to infer the parameter of interest.
As far as telling from a single test if a person tagged as Positive
is really infected, we have seen that the prior `assumed proportion' of infected
individuals in the entire populations plays a major role.
We have seen how to get the probability of interest reasoning
on the fraction of positives really infected in the sample
of positives. In more general terms
this probability has to be calculated using
Bayes' theorem, that will be shortly reminded in the next section.