So I’m gradually approaching the exam segment of the second year of my pursuit of an MSc in Epidemiology. It feels like someone has been speaking Greek to me for months (otherwise known as EPM202 and EPM304) and – all of a sudden, I can understand (at least a little more) what is being said.
How do you decide on the best strategy for answering a research question?
Newsflash: It depends on what you want to know!
I know this will come as a shock to all those epidemiologists out there…But that phrase the tutors have been patiently repeating for months is (deep breath)…true.
If you are a public health official trying to figure out how many people in your region may come down with, say, Awful Disease X, what you care about is how well you can PREDICT that number. You are interested in (drum roll) independent risk factors in general. You don’t much care whether Awful Disease X (or ADX to those in the know) is caused by, say, working late on weekends, but you sure do care whether people who spend most of their free time at the office have higher odds of getting the illness.
If on the other hand you are a researcher trying to find a way to reduce incidence of ADX, it may not help you much to know that weekend workers have a greater tendency to succumb. What you care about is understanding the CAUSES of ADX in your region. Let’s hear it for … causal factors and their uninvited but always-pushing-in cousins, The Confounders (sounds like a bit like a rock band…). Researchers want the purest picture of any relationship between possible causes and Awful Disease X, so they are less interested in any risk factors other than the bothersome confounders, which must be done away with via the mysterious process known as “controlling for them,” leaving behind (in theory) a nice clean association (or not).
The bottom line is that, depending on who you are and what you want to know about a particular health outcome, you will be interested in different types of variables, analyze their relationships differently, and choose to highlight different points in your conclusions.
It strikes me this is a lot like the rest of life. Getting the strategy of analysis right depends on what you will use your results for—just as the “right way” to do anything depends on what you want to do. What should I study? Where should I live? How should I earn a living? More serious questions too: Whom should I follow or lead? What should I endure and what should I fight against?
By chance, the daily “Moment of Happiness” quote (see Gretchen Rubin’s The Happiness Project) this morning was from Thomas Merton, who pointed out that “Thinking about monastic ideals is not the same as living up to them, but at any rate such thinking has an important place in a monk’s life, because you cannot begin to do anything unless you have some idea what you are trying to do.”
Epidemiologists may not all be monks, but Thomas has a point.