Halfway through the 2021/22 school year, some Massachusetts school districts dropped their mask requirements. This created a natural experiment.
We estimated: The causal effect of removing mask mandates in those districts on COVID levels
"What did we find?"
School districts which removed mask policies had an immediate increase in COVID cases in both students and staff.
This increase in cases got bigger over the next 10 weeks.
"How much did removing mask policies matter?"
Schools which removed mask policies had an average *increase* of 44.9 cases per 1,000 people over a 15 week period compared to if they had not removed their mask polices.
This adds up to an estimated 11,901 *extra* COVID cases!
Implication: No mask requirement means kids miss school!
We estimated that students had an average increase of 39.9 cases per 1,000.
If all students who tested positive followed state guidelines to isolate for 5+ days, that's almost 20,000 *extra* missed school days!
Implication: No mask requirement means teachers & school staff miss work!
For all staff, we an estimated 81.7 extra cases per 1,000!!
That's how many *more* teachers & staff got sick with COVID than would have if those school districts had kept mask requirements in place.
"What about alternate explanations?"
School districts that dropped mask requirements tended to have better budgets, better buildings, better vaccination rates, and lower community COVID levels.
Our study shows those weren't enough to make dropping mask mandates work!
"Why should you believe our study?"
This study is not a randomized controlled trial, it's an observational study, but it *can* still tell us about cause and effect.
For details on why & how the approach we used works, see this earlier thread:
Eastern Massachusetts school districts which removed mask requirements in March 2022 saw a *dramatic increase* in COVID cases compared to if those districts had retained mask requirements.
And this was *despite* all other mitigation efforts those districts did.
Here's a link to the pre-print for those that want to delve more deeply into the science: medrxiv.org/content/10.110…
Adding to say: First author Dr Tori Cowger’s twitter is @ToriCowger (which I probably should have guessed😆🤦🏼♀️)
One last thank you: to #econtwitter who do an excellent job of summarizing & tracking advancements in methods, and helped me stay up to date on new & improved DiD estimators over the past year.
A few people are critiquing this study saying that getting rid of mask policies simply means more kids were tested for COVID. I disagree, because we didnt just look at contact tracing.
But even if were a little true this critique actually strengthens the evidence for mandates!
If, as these critics themselves are claiming, lifting mask mandates means more kids are tested and so more kids test positive, that STILL means lifting mandates causes more missed days of school — even if it only changes detection not transmission!
So, no matter whether you believe our interpretation of the data that there are more *total* cases without mask mandates, or you believe the counter-interpretation that there are more *detected* cases, you still have to end up at mask mandates keep kids in school!
The point here being that, no I don’t think there’s any reason to suspect detection is what changed, and also that any critique which relies on everything about our study being correct except the final conclusion, is a pretty weak criticism 🤷🏼♀️
Some people are asking why we didn’t control for “obvious” confounders.
The simple answer is: we did. The method we used implicitly controls for all of the differences between districts that don’t change over time.
Why should you believe me?
My research expertise is specifically in how to control for confounding (and other biases) in health research so we can get good answers to clear questions.
I did my doctorate on this at Harvard and I work on it every day at Boston University.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
Today's #tweetorial is all about a tool called difference-in-difference analysis.
Economists & epidemiologists both use this tool, and we affectionately call it diff-in-diff or DiD since the real name is pretty long! I'm going to use DiD to save character space.
So, what is DiD? The simple answer: a way of comparing how places change over time.
We collect data on how something changes over time from Place A, e.g., COVID cases. We do the same in Place B. And then we compare the changes.
I’ve noticed that Americans dont really believe in the concept of prevention.
When good advice leads to NO disaster, people think they’ve been fooled rather than that the disaster was averted.
People worry about the Boy Who Cried Wolf, but IMO they get the moral all wrong.
1/
For those who don’t know, The Boy Who Cried Wolf is a short tale (fable) about a bored little shepherd boy who cries out about fake wolves, brings the villagers running to help, and then laughs at their supposed foolishness in being duped.
But when the wolf does finally appear, the boy calls for help and no one comes because they’ve learned he’s a liar.
On realizing he’s not back for the night, they traipse up to his pastures & discover that his flock is scattered, scared by the wolf, & the boy is sobbing.
With @DrCorlin & Amanda Sullivan, we tried to answer the question “how much epidemiological research published in medical journals actually includes epidemiological expertise?”
First, we selected 15 top journals (based on Scimago rankings) from the categories “epidemiology”, “general medicine”, and “specialty medicine” (5 in each)
For each one, we randomly chose 3 issues from 2000 & 3 from 2010, plus the first available issue in 2020.
Next, we (and by “we” I mean Amanda who did the data collection!) compiled a list of first, last, & corresponding authors of all original research articles in the selected issues.
People in my mentions complaining fridges aren’t anything like masks because “fridges are harmless” and oh boy do I have news for them 😆
They’ve got it backwards: Masks aren’t going to suffocate you. But fridges might!
Come down an internet rabbit hole about fridges with me
“refrigerator death” has a whole wikipedia page
This is when you get trapped in a fridge & suffocate because they are air-tight. Modern fridges have safety mechanisms but maybe just dont get inside one? en.m.wikipedia.org/wiki/Refrigera…