Representative vs. homogenous samples

[basophilic]

For a research experiment, in what situations would it be better to have a representative sample vs. having a homogeneous sample?

My guess is this: representative samples help improve the external validity of the study but not internal validity (i.e. causality); but homogenous samples increase internal validity but not external validity. The examples that pop in my mind would be correlational studies on say socioeconomic status and schizophrenia frequency for the former; causal studies (most biomedical, biological, neuroscience studies) favor homogenous samples to be able to control for external confounding variables so that the causal relationship can be strengthened. Is this a fair statement to make IN GENERAL?

---

Members don't see this ad.

 

[basophilic]

Anyone have any input?

 

[bee17]

Your explanation sounds right to me. A homogenous sample is good because it allows you to control external variables that might interfere with your experiment. A representative sample is good because it gives you a good representation of a population.

 

[basophilic]

Your explanation sounds right to me. A homogenous sample is good because it allows you to control external variables that might interfere with your experiment. A representative sample is good because it gives you a good representation of a population.

And hence for homogenous samples: you gain better internal validity at the expense of external validity.
And for representative samples: you gain better external validity at the expense of internal validity.
Correct?

 

[bee17]

Generally, yes. Unless the population you are testing happens to be very homogenous, in which case a representative sample would also be a homogenous one.

 

[basophilic]

Generally, yes. Unless the population you are testing happens to be very homogenous, in which case a representative sample would also be a homogenous one.

Ah ok, thanks for the help.