2 - Continuous Distributions: Confidence Limits

Summary

A population represented by a sample

The populations about which we wish to make statements and draw conclusions are represented in biological surveys and experiments by samples (see § 1.1). These samples consist of individuals. In some investigations these individuals are whole plants or animals but more generally they range from individual cells, or even organelles, to plots of forest trees. They can take the form of organs, tissue preparations, extracts, and even environmental locations. Despite this diversity they have in common the fact that they contribute an item of information relating to one or more of their attributes. For the moment we are concerned only with situations in which there is information relating to one kind of attribute and in which the information consists of measurements on a continuous scale, such as weight, volume, area, length, concentration, rate, pH, etc.

Suppose that as part of an autecological study of the bracken fern (Pteridium aquilinum) we wished to assess the performance of the fern within a certain area, Area 1. Frond (leaf) length could be included among the performance parameters. There are thousands of fronds in the population of the area and they vary conspicuously in length. We might decide to measure a sample of 100 fronds. Clearly the sample should be representative and must therefore be selected without bias. This is more difficult to achieve than it might appear, simply measuring one frond here and another there in an irregular manner will not do.