The goal in science is often to try to estimate some parameter of interest. We sample because it is often either impractical (due to time or financial constraints) or impossible to measure every individual of a population of interest. If sampling is done correctly, we hope that our estimate of the parameter can be generalized back to the entire population of interest.
Types of Experiments
- Monitoring – the researcher does not manipulate a variable. S/he records data on what is happening in the environment, usually on experimental units that differ in one or more ways.
- Experiment – one or more variables are deliberately manipulated, and experimental units are randomly assigned to these treatments using one of the sampling techniques described below
- Quasi-experiment - similar to an experiment in that one or more variables are deliberately manipulated. However, there is either no random assignment of experimental units to treatments or no control group. Quasi-experiments can be common in ecological research.
To ensure that we obtain a sufficient sample size and a representative sample size we use replication. A rule of thumb for replication is to sample a minimum of 10% of the area or population to which you are studying.
Bias occurs when members of the population of interest do not have equal chances of being chosen for the sample. This may result in a biased estimate of the parameter of interest. Standard sampling techniques for minimizing bias include: