Understanding the sources of measurement errors
Statistical errors are due to the very nature of the sampling process. They are unavoidable but can be accurately estimated, which enables us to take them into account.
Consider a simple situlation. You have a culture of bacteria. There is 1 bacterium per ml. To measure bacterial concentration, you remove 0.1 ml from the culture. It is not possible to have 0.1 bacteria per volume, the number must be a whole number - either 0 or 1.
What is probability that you will have exactly 1 bacterium within this volume? The answer is that on average, you will have 1 bacterium in your sample once out of every 10 sampling events. If you calculated bacteria concentration based on a single sampling you would always be wrong! This is, however, rarely the case. Normally, there are many more bacteria/ml. Nevertheless, the same logic applies.
The probability that your sample will accurately reflect the concentration of bacteria in the original solution is described by what is known as the Poisson distribution. As the concentration of the objects increases, the sampling error decreases as the square root and the magnitude of the error compared to the true number decreases. (TOM - HELP).
Want to know more about the Poisson distribution and statistics? - ask your vTA.