The world is a very complex place. Even in the relatively controlled world of B2B, there are always a huge variety of factors that exert some sort of influence on whatever you might be interested in understanding. Many of the factors that influence how people and businesses behave are random and/or hidden from our view. As a result, it is impossible to account for all – or sometimes even most – of the factors that influence outcomes we are interested in.
The variables that make up these hidden influences are called background confounders. They are a type of independent variable that is not of particular interest to your study, but which affects the relationship between variables you are interested in. It’s something that can sway your understanding of how one thing affects another, but isn’t something you care very much about at present.
While many of them are impossible to account for, we can account for some of them. When researching, you want to account for, or control for, the presence of background confounders so you can hone in precisely on the variables of interest.
For instance, let’s say we’re trying to figure out if a company’s size affects how quickly they make purchase decisions. Currently, the focus is not on how factors like industry or type of solution affect the timing of decisions. We just want to get the impact of company size.
You cannot simply ignore the fact that companies that make purchases exist within industries and sell solutions of different types. One way to do the research would be to limit your sample to buyers from just one industry and which sell just one type of product. That, however, might make getting an adequate sample difficult. And, you would not be able to generalize your results outside of that industry and product type.
The solution is to mathematically control for the background confounders, in this case the type of solution and industry of the buyer. Doing so removes the influence of those factors and yields a result that can be generalized across multiple industries and solution types.
When discussing the result of an analysis where you have controlled for background confounders, you would describe your findings by saying that you controlled for, or held constant, the background confounders.
