Bivariate data: population or sample?
In most bivariate data sets, we have no interest in the specific individuals from which the data are collected. The individuals are 'representative' of a larger population or process, and our main interest is in this underlying population.
Example
Data were collected by biologists from 15 lakes in central Ontario to assess how zinc concentrations in an aquatic plant were related to zinc concentrations in the lake sediment.
The biologists want to generalise from these specific lakes (and sediment samples) to describe the relationship between zinc concentrations in sediments and plants in a way that might be used to predict plant zinc from sediment samples in other similar lakes.