The goal of unbiased stereology is making accurate estimates of expected values for parameters through accurate and efficient sampling.

Systematic-random sampling

The above figure shows an example of Systematsi-Uniform-Random sampling (systematic-random sampling for short). The sampling locations vary in a Systematic manner; the distance between the samples is Uniform; and, the first location in the tissue is Random.

An alternative to systematic-random sampling is pure random sampling: Select random location for first sample, make first estimate; select random location for the second sample, make second estimate, and so on.

Both systematic-random and pure random sampling are unbiased

Both systematic-random and pure random sampling yield estimates that converge to the expected parameter. However, estimates with systematic-random sampling converge to the expected value sooner. Systematic-random sampling makes more precise estimates in less time; therefore, systematic-random sampling is more efficient than pure random sampling.

The figure below shows actual data from a study of the volume of a defined region (Vctx, in cubic centimeters) for tissue sampled by systematic-random (systematic) and pure random (random) approaches. The same reference space was sampled multiple times using either systematic or pure random sampling. The results of the parameter estimate for each sample are shown on the y axis. E(Vctx) refers to the expected value for the parameter.

The above figure shows that for systematic-random sampling the parameter estimate converged on the expected value after only two samples. After six samples the estimate from pure random sampling was still centered a clear distance above the expected value.

next page: Importance of Sufficient Sampling >>