In statistics, it's always impossible to do experimentation on the entire population. Sample of this population are then used to estimate characteristic of this population (ie the statistics).
A sample statistics is an estimate of a population parameter
Because of this estimation, we get what's called sampling error.