model tag, it is written that there is a distinct element of $P$ (probability) which generates the observed data. Though statistical model involves a random error term ($\epsilon$), will I be able to generate the same observed data for a certain probability ?
And in Wikipedia, it is written that
A statistical model embodies a set of assumptions concerning the generation of the observed data, and similar data from a larger population.
When should i get the observed data and when should i get similar data ? For different probabilities or only one probability can generates different set of data , as statistical model involves random error term ?
Can i see it empirically ?