TABLE OF CONTENTS As discussed in Chapter 1, we would like to obtain samples that are representative of the entire lot. That way we will know exactly (or at least within the analytical measurement) the values of the properties of interest. From our experience, however, we know that these values vary from sample to sample. Thus, the samples cannot all be representative. How can we get representative samples, or at least approximately representative samples?
The key is to apply the statistical principle of random sampling. We saw in Chapter 2 that the sampling of individual units at random minimized the sample-to-sample variation. Random samples may seem to be taken in an arbitrary and unorganized fashion. But, in fact, they are more representative, and consistently so, in repeated sampling situations (day in and day out) than samples obtained any other way. Actually, we all practice random sampling in simple situations, as we will illustrate below. The real trick is applying this principle to solids, liquids, and gases, where individual units are rarely available for selection.
