The importance of sufficient sampling

Good stereology requires a sufficient intensity of sampling to capture the majority of biological variation in a parameter. If we properly apply unbiased sampling and estimation techniques, but fail to adequately sample the tissue of interest, there is a risk of basing conclusions on accurate, but imprecise data.

As illustrated by the U.S. Presidential election in November-December 2000, the results of sampling total number of votes for each party's candidate vary depending on how intense the population is sampled. The closer the two values under comparison, the greater role is played by the intensity of sampling.

When large differences are present, high levels of sampling error have little effect on the outcome. As the true difference in votes for number of votes decreases, sampling error plays an increasingly important role.

The power of unbiased sampling in stereology allows users to estimate millions or even trillions of objects by counting less than two hundred objects in a defined tissue space.

With such an amplification scheme, an innocent identification error there and assumption in the methodology there, in conjunction with inadequate sampling, can easily lead to an inaccurate estimate of the expected value.

Another reason for thorough sampling is to reduce the effects of a single error. Stereological data collection, like vote counting, is done by humans, and we can all make an error in the best of conditions.

However, good stereology is efficient stereology. Over-sampling beyond the point of diminishing returns it is not good use of resources. As noted by the esteemed Swiss stereologist, Professor Ewald Weibel, the most rational approach is: Do More Less Well. That is, optimize sampling to achieve stable estimates with minimal time, effort, and resources.

In summary, adequate and unbiased sampling is part of the sterelogist's commitment to make accurate, precise, and unbiased estimates.