The Tyranny of the Mean

In the real world variability is a fact of life.  Every individual is unique. Yet when doing biological research we often try to minimize these differences in order to compare one group of individuals to another. For example if you wanted to know whether a particular drug is effective in treating a disease you would compare a group of individuals who received a medical intervention to another group of individuals who received a placebo. However every individual's genetics, environment, and disease progression is likely to be different resulting natural variability in the way that they respond to the treatment. In order to have confidence in the results of the study you must control for as many factors as you possibly can, randomly assign individuals to groups and have as many individuals in your study as possible.  When reporting results we often take the average (mean) of measurements for each group. Statistics tell us what the probability is that random chance could account for the differences observed between the two groups.  Large sample sizes are the best way to increase your statistical power (chance of seeing an effect) however we must remember that these results come with a caveat.  The mean results seen will not be true for every individual.

Picture taken by Marie Suver

 A talk I recently saw by Eve Marder from Brandeis University highlighted how important individual variability can be for biological systems.  The Marder lab studies the Somatogastric Ganglia (STG) of the crab.  The STG is a bundle of neurons that control the muscles of the gut that grind food. The STG is what is called a central pattern generator, neurons in a central pattern generator produce rhythmic patterns of firing (action potentials) without input from the sensory system in order to drive repetitive motor patterns. What's great about this preparation is that it can be removed from the animal and the neurons still maintain their characteristic pattern of activity.   In my previous post I talked about the fact that action potentials are generated by ions flowing across the cell membrane through ion channels.  The STG consists of a small number of large, easily identified neurons (see photo) making it relatively easy to stick an electrode inside of the cell to measure changes in the voltage that occur when ions flow into or out of the cell.  This is called ion conductance.  Each type of neuron has a characteristic pattern of firing action potentials but the way that this pattern is generated is different from individual to individual.  What is  most surprising was that in some cases the mean measurement of a particular type of ion conductance across a population actually occurred in very few of the individuals tested.  Instead there were several different ways ionic conductances could be balanced to generate the rhythmic spiking activity of the neurons and the measurements tended to cluster around these values.  This result is an important reminder that we cannot always take a mean value at face value. In a previous post I talked about how frustrating it is when scientific studies are not as clear-cut as we had hoped.  Misinterpretation of variation is one factor that can contribute to an incomplete understanding of our results.  One thing is for sure: we need more data!