NOTES
Here's how to apply the idea of independent data to clusters - pick one subset of the total data set to come up with a theory - like "the incinerator is causing increased risk of disease" - you then have many more subsets to use for testing the idea.