Thank you! I like the term you coined, "implied causality", and its classification into 4 types. How would you remedy the misleading causality for each? Is there a way to make a strong claim of causality with data visualization?
The problem resides in the data. Unless the data come from a controlled experiment (or a natural experiment), causality is inherently uncertain. Most data is what scientists call "observational data." While causality cannot be directly inferred from observational data, different degrees of plausibility can be derived from it. One of my favorite guides is the Bradford Hill criteria, available at https://en.wikipedia.org/wiki/Bradford_Hill_criteria. These criteria make causality more likely/plausible. How these criteria can be used when designing data visualizations is an open question. I'll try to post some ideas in this regard. P.S. I am going to talk about these ideas in my upcoming webinar: https://maven.com/p/d1a4ec/don-t-get-fooled-avoid-the-illusion-of-causality-in-charts?utm_medium=ll_share_link&utm_source=instructor. I hope you'll join me!
Visualizing implied causality is one of the chief tools in executive management, public policy development, and misinformation proliferation. One of the most important skills in those fields is the crafting of language to legitimize the apparent relationship on the chart.
Weaponized visualization (aka the chart effect) is so effective because it exploits the desire of most people for easy answers that reinforce preconceived notions. Short of analyzing the data oneself, the most effective way to minimize exposure to the chart effect is through the cultivation of reliable and trustworthy sources, a tricky endeavor in itself.
Rather than minimizing exposure, I prefer the idea of empowering people with critical data thinking skills so that they can catch problems and avoid absorbing the message as is. What do you think?
I see this a lot in practice.
What I see a lot is people looking for any explanation for why a number moved.
When one is found the job is done.
Whether that explanation is a/the cause doesn’t matter - because the uncertainty is gone - so all’s well again.
Yep
A good reminder to consider all the factors in play - not just what you're testing for. Thanks for sharing!
Thanks for reading and commenting!
Thank you! I like the term you coined, "implied causality", and its classification into 4 types. How would you remedy the misleading causality for each? Is there a way to make a strong claim of causality with data visualization?
The problem resides in the data. Unless the data come from a controlled experiment (or a natural experiment), causality is inherently uncertain. Most data is what scientists call "observational data." While causality cannot be directly inferred from observational data, different degrees of plausibility can be derived from it. One of my favorite guides is the Bradford Hill criteria, available at https://en.wikipedia.org/wiki/Bradford_Hill_criteria. These criteria make causality more likely/plausible. How these criteria can be used when designing data visualizations is an open question. I'll try to post some ideas in this regard. P.S. I am going to talk about these ideas in my upcoming webinar: https://maven.com/p/d1a4ec/don-t-get-fooled-avoid-the-illusion-of-causality-in-charts?utm_medium=ll_share_link&utm_source=instructor. I hope you'll join me!
Visualizing implied causality is one of the chief tools in executive management, public policy development, and misinformation proliferation. One of the most important skills in those fields is the crafting of language to legitimize the apparent relationship on the chart.
Weaponized visualization (aka the chart effect) is so effective because it exploits the desire of most people for easy answers that reinforce preconceived notions. Short of analyzing the data oneself, the most effective way to minimize exposure to the chart effect is through the cultivation of reliable and trustworthy sources, a tricky endeavor in itself.
Rather than minimizing exposure, I prefer the idea of empowering people with critical data thinking skills so that they can catch problems and avoid absorbing the message as is. What do you think?
Awesome, I have learned so much with this!
Happy to hear that Victoria!