Impact of Variance

NOTE: The article below touches on statistics associated with COVID-19. I do not claim to be a public health expert, nor am I a public health statistician. Please take the below in that context, and please seek out your information on this or any public health situation with your local/national public health agency.

Recently, I found myself in two separate instances looking at the impact of variance (note: I’ve interchangeably called this impact of error and impact of change).

“Impact of Variance”:

We use data to draw insights and conclusions about decisions to make. However we often seem to look at the quality of data as having equivalent impact. When looking at user collected data, what is the impact of an error in the data.

For example, In my work setting, we recently conducted an analysis to understand factors affecting operational performance. Aside from the debate between leading and lagging indicators, we found ourselves discussing the impact of variables that did not score highly on overall reliability. We knew this to be true as there are multiple factors affecting performance, and none could singularly be the primary driver (note this is still work in progress). However, one of the debates that emerged was around the relative impact that one variable had over another. Both were discrete integer data (i.e. 1, 2, 3, etc…), and both scored relatively the same in our model. However, what we knew to be true was that both the marginal impact of one was dramatically different in the…