Read on Twitter
The scientific method is one way of gaining knowledge about the world. Most of us learn in school about how this works for experimental lab research, but I think very few of us learn how it works if we can’t experiment.
So let’s talk about how to get knowledge from observations.
So let’s talk about how to get knowledge from observations.
Observational data is what we call it when we collect information from the world around us, without our having done anything intentional to affect the process of interest.
(An important side note I’m not going to discuss, is that sometimes just observing things changes them)
(An important side note I’m not going to discuss, is that sometimes just observing things changes them)
One of the strange things about learning from observational data is that this is something we all think we’ve known how to do since birth, but also something few of us really know how to do.
How did you learn to walk, talk, eat: by observing other people.....
How did you learn to walk, talk, eat: by observing other people.....
... but even tho we learn things by observation all the time, we seldom learn *only* by observation.
Have you ever seen a baby get up and walk across the room on the first try? No, because observation alone isn’t enough. We usually have to experiment, even if just on ourselves.
Have you ever seen a baby get up and walk across the room on the first try? No, because observation alone isn’t enough. We usually have to experiment, even if just on ourselves.
The problem is, there are many many scientific questions where experimentation is difficult, impossible, or unconscionable.
For example, we can’t experimentally measure the SARS-CoV-2 infection mortality rate or the effect of smoking initiation on lung cancer.
For example, we can’t experimentally measure the SARS-CoV-2 infection mortality rate or the effect of smoking initiation on lung cancer.
For these types of questions we can’t do an experiment to test and validate our conclusions, so we need another framework for evaluating whether we are observing the right data and drawing the right conclusions from it.
Luckily, obtaining validated knowledge about people, systems, processes, and causal effects from observational data is what the field of epidemiology was created to do! It’s our bread & butter.
Even more lucky for you, we love to nerd out about even the smallest details
https://abs.twimg.com/emoji/v2/... draggable="false" alt="😆" title="Smiling face with open mouth and tightly-closed eyes" aria-label="Emoji: Smiling face with open mouth and tightly-closed eyes">
Even more lucky for you, we love to nerd out about even the smallest details
Okay, so that’s why you should listen, but what you really came here for was how to do it.
How do we get (scientific) knowledge from observational data?
How do we get (scientific) knowledge from observational data?
Step 1: define‘knowledge’
In science, we usually mean we observed something in a reproducible way *and* if it were NOT true our process for observing it would tell us we were wrong.
The last part is key! Without a way to know we’re wrong, we can’t be confident we’re right!
In science, we usually mean we observed something in a reproducible way *and* if it were NOT true our process for observing it would tell us we were wrong.
The last part is key! Without a way to know we’re wrong, we can’t be confident we’re right!
Using this definition, we can come up with a basic plan for gaining knowledge from observational data:
Make an observation, think of every possible way it could have happened real or unreal, & systematically start eliminating the wrong ones; what you’re left with might be right.
Make an observation, think of every possible way it could have happened real or unreal, & systematically start eliminating the wrong ones; what you’re left with might be right.
Most of what we learn from observational data starts out as an idea formed by someone seeing something weird. This might be written about in a ‘case study’ or ‘case series’.
For example, @MonaHannaA noticed strange symptoms shared in common among children in Flint Michigan.
For example, @MonaHannaA noticed strange symptoms shared in common among children in Flint Michigan.
For some reason my thread seems to have split in half. It continues here: https://twitter.com/epiellie/status/1250907076966133760?s=21">https://twitter.com/epiellie/... https://twitter.com/EpiEllie/status/1250907076966133760">https://twitter.com/EpiEllie/...
