Discover and read the best of Twitter Threads about #statstwitter

Most recents (20)

There seems to be growing interest in the scientific and ethical merits of using an adaptive or platform design for clinical trials, particularly in the setting of large scale #COVID19 international collaboration. Are there any drawbacks?
It depends on the stopping rules, but halting poorly performing arms can bias the summary estimate away from the null. This bias can be magnified with publication bias in systematic reviews as described in @JAMA_current here jamanetwork.com/journals/jama/… #statstwitter
The problem is some variation is random, so unusual benefits or harms could be due to chance alone. The idea of "regression to the mean" is where unusual results become less unusual over time as the effect of random outliers is diminished over time. bmcmedicine.biomedcentral.com/articles/10.11…
Read 8 tweets
This thread racked up over 34K retweets. It's an expertly written thread that makes many important points. But its central premise – that incorrectly pooling data is the key problem – is at best misleading and at worst wrong. #epitwitter #statstwitter 1/13
It's an insidious example of armchair epidemiology: “The experts can't (fully) explain what's going on, but I can. It's simple. The real insight is from [waves hands] [insert cool paradox here].” Let me explain why. 2/13
I'm a data scientist, health economist and public health policy professor. My research uses rigorous statistics methods to draw accurate conclusions from non-random-sample data. Much of my work develops data-driven policy for another deadly respiratory disease: #tuberculosis.3/13
Read 14 tweets
Was asked for personal favorite resources for improving methods and statistics skills. I promised to make it a thread, so here it is

1/n
I work in medical research, so that is going to be my focus here too. But I’d like to think the resources are relevant to a wider audience

This list should not be taken as a guide to become a statistician, nor is it a must-read list for all academics (obviously)

2/n
My personal view is that medical research would benefit from involving trained statisticians earlier and more frequently; not from everyone trying to become one

Here are some good arguments by @statsepi:
medium.com/@darren_dahly/…

And some more: medium.com/@darren_dahly/…

3/n
Read 20 tweets
1. Why do some #COVID19 cases have severe outcomes while others are mild? Many of the comorbidities associated with poor outcomes are also associated with #obesity. A new paper looks even deeper to find an underlying link ==> chronic inflammation nature.com/articles/s4136…
2. The authors of this paper are familiar with how one size does not fit all with obesity. @DrSharma established the Edmonton Obesity Staging System to help characterize obesity beyond size alone to address physical, mental, and functional health. drsharma.ca/wp-content/upl…
3. The authors identify age-dependent defects in T-cell and B-cell function and the excess production of type 2 cytokines could lead to a deficiency in control of viral replication and more prolonged pro-inflammatory responses, potentially leading to poor outcome
Read 11 tweets
This piece on #SARSCoV2 viral load in children and adults has been highly influential.
Can you extrapolate from “Data on viral load” that “we have to caution against an unlimited re-opening of schools and kindergartens in the present situation”
I have serious concerns. 1/7 Response to Christian Drosten group report of viral load by age group https://twitter.com/c_drosten/status/1255555995671150597?s=20
There are two issues: science and politics
On the science
1. There is no methods section about how study population was selected and who they represent
– yes, I know it is a bunch of samples tested in a virology lab - with no denominators about how many samples tested by age 2/7
2. The very low number of samples from children already says a lot about selection into the study. There is no information about their clinical characteristics, stage of infection, etc.
And unequal numbers across the groups makes them very difficult to compare, even visually 3/7
Read 7 tweets
Who’s up for a #tweetorial about adherence, per-protocol effects, and randomized trials?

@_MiguelHernan and I have a new paper out in collaboration with the #CHARM trial team. Will this be the one that finally convinces you? 😄

OA Link: authors.elsevier.com/a/1aUFF_ZI2ghd…
First up, what are per-protocol effects and how are they different from intention-to-treat effects?
The per-protocol effect is the difference in the outcome that we would see if everyone in our trial had followed treatment protocol A compared to if everyone had followed treatment protocol B (regardless of what protocol people did or didn’t actually follow!)
Read 20 tweets
Hello #epitwitter and #statstwitter,
There was a recent discussion about the Hosmer-Lemeshow goodness of fit test. I thought it would be interesting to talk to Dr. Lemeshow (who is not on twitter) about his thoughts on the test. This thread has some highlights from our chat. 1/n
In the late 70’s, Hosmer and Lemeshow were struggling with the question “How do you know that the probabilities produced by logistic regression models reflect reality?” This was the motivation for developing the Hosmer-Lemeshow goodness of fit test. 2/n
As with a linear regression, we can’t only look at the estimates from the model to know if the fit is good. For linear regression we can look at a plot (e.g., residual plot) to assess model fit. 3/n
Read 15 tweets
Starting 2020 off right by getting the benefit of hindsight & reading some classic papers.

#epielliereads
1st up, Take the Con out of Econometrics by Edward Leamer, 1983: all about pros & cons of randomized & observational studies plus answers to many of recurrent #econtwitter, #statstwitter, & #epitwitter arguments & even why #ML can’t do causal inference!
jstor.org/stable/1803924
In Part I randomized experiments vs natural experiments, aka observational studies, Leamer provides a nice summary of some of the problems that can arise in RCTs including that randomization guarantees validity *on average* but that chance imbalance can make any given RCT biased
Read 13 tweets
This my *top 10* of favorite methods papers of 2019

Appearing in a single thread and in random order
Disclaimer: this top 10 is just personal opinion; I’m biased towards explanatory methods and statistics articles relevant to health research
#1: A plea against dichotomization of study results as statistically significant or not. With three authors, 800+ signatories and #2 Altmetric score of the year, this article belongs on the list. Agree with the message or not, or only a little

go.nature.com/2Qd0eUI
Read 12 tweets
JOB ALERT: if you are an aspiring trial statistician (or an experienced trial statistician seeking a change of scenery) that would consider a move to Pittsburgh, feel free to DM me for details.
Background in stats, biostats, epi, or any related field, but clinical trial experience (student experience counts) is highly preferred. The group doing the hiring is specifically looking for someone with interest in trials.
Tossing this out to #statstwitter and #epitwitter for a signal boost.
Read 3 tweets
THE OBESITY PARADOX IS NOT A PARADOX

A tweetstorm by a frustrated epidemiologist

I was disappointed to see an "obesity paradox" article in @AmJEpi

Instead writing a letter to the editor, I decided that Twitter is a better way to reach ppl on #epitwitter #medtwitter

1/
Before I begin: I have the utmost respect for @easchisterman and the @AmJEpi team. But, I have a particularly strong reaction to articles claiming to have evidence of a “true” obesity paradox.

Science moves forward through scholarly debate. Let’s keep the discussion courteous!
Also, PLEASE RT and share with colleagues on #epitwitter #medtwitter #statstwitter #academictwitter

I'd love to engage and answer any questions you may have!

Here's the article I'm tweeting about:
ncbi.nlm.nih.gov/pubmed/31504124
Read 11 tweets
"Between 2006 and 2015, ... colorectal cancer rates increased by 3.47 percent among Canadian men under age 50. And from 2010 to 2015, rates increased by 4.45 percent among women under age 50."

Let's check the data ...
1/2

nyti.ms/2Yyk9zx
Who teaches that fitting a regression line is drawing a straight line and drawing as many other straight lines as needed and wherever needed?
2/2

@f2harrell @statsepi @MaartenvSmeden #epitwitter #statstwitter #peerreviewed @JAMANetworkOpen
jamanetwork.com/journals/jaman…
I am adding a reference to the method here, which I didn't know about (and which I still don't understand why that can be applied to 45 datapoints without over fitting--maybe someone can explain).
Read 3 tweets
@statsepi @ashtroid22 @Epi_D_Nique @healthstatsdude @stephensenn (1/n) The problem Senn is referring to here, was solved by Mindel C. Sheps in NEJM in 1958. I proposed a counterfactual causal model that formalizes the intuition behind her argument in this paper: degruyter.com/view/j/em.2018…
@statsepi @ashtroid22 @Epi_D_Nique @healthstatsdude @stephensenn (2/n) Consider this example: If people both Russia and Norway are randomized to play Russian Roulette at the end of every year, and the baseline risk of death in a year i 0.5% in Norway and 1.0% in Russia: Calculate how many people will die in the active arm in each country.
@statsepi @ashtroid22 @Epi_D_Nique @healthstatsdude @stephensenn (3/n) If you do the maths, you will find that 17.5% die in the active arm in Russia, and 17.1% die in the active arm in Norway. This means risk ratio is 17.5 in Russia and 35 in Norway. Odds ratio is 21 in Russia and 41 in Norawy, and survival ratio is 5/6 in both countries
Read 10 tweets
Cool moment today re: academic power of #statstwitter: looking through the reference list for a paper that I'm working on and seeing so many folks that I've 'met' through Twitter on the list of papers cited...
Read 3 tweets
A few people have DMed me about career advice as in "How do I get started in statistics?" I'm not where I want to be in life yet! But I have switched fields *multiple* times. So here are my tips on moving from insider to outsider ...
#epitwitter #statstwitter #datascience
1. Google a few people who started where you are and got to where you want to be. Look at their resume, work backwards and copy what they did. (This doesn't always work, especially if you are the first person like you.)
2. Find a person who's where you want to be. Tell them your background honestly. If they tell you that you aren't suited to their industry, ignore that for now. Get them to name one concrete gap that you have. Fill the gap. If you don't know how, then ask. Repeat as needed.
Read 9 tweets
Hello all #dataviz fanatics! It’s time to kick off our #datavizbook #epibookclub for @kjhealy’s new book, Data Visualization.

Remember, if your book hasn’t arrived yet or you’re waiting on the library, you can read it here: socviz.co

#epitwitter #datascience
This week, we’re reading the Preface and Chapter 1: Look at Data.

I’ll post some highlights from each and then I hope you’ll chime in with your thoughts, comments, questions, etc.
In the Preface we get a nice overview of the goal of this book: the why and how of good data visualization for beginners, including practical applications in R with ggplot2.

The book doesn’t assume any prior knowledge of R, & covers everything #dataviz from scatterplots to maps.
Read 18 tweets
Are you still shaking off the holiday? I know I am!

How about a #cartooncausalinference #tweetorial about casual graphs to ease us into the new year?

#epitwitter #DAGsfordocs #FOAMed #MedEd #statstwitter #econtwitter
The most common type of causal graph (at least on #epitwitter) is the directed acyclic graph, or #DAG.

DAGs have two main components: variables (also called nodes), and arrows (also called edges).

In the DAG below, there are 3 variables: sleeping, Santa, and presents.
The variables are ordered based on time — you have to go to sleep before Santa can come to your house & then he’ll leave presents!

Causation and time both flow in the direction of the arrows.
Read 24 tweets
🎄🎅🎄🎅🎄🎅🎄🎅🎄🎅🎄🎅🎄🎅🎄🎅🎄🎅🎄🎅🎄

I’m feeling festive! I want to spread the cheer this holiday season by giving the #GiftOfCode for making beautiful figures! #EpiTwitter #StatsTwitter

🎄🎅🎄🎅🎄🎅🎄🎅🎄🎅🎄🎅🎄🎅🎄🎅🎄🎅🎄🎅
When I was starting out, I remember how it would take HOURS to get my figures looking just right.

Nothing was more valuable than example code for learning different #DataViz options 🎁
First up:

These options in #stata for making this graph presentation-ready!

adjustrcspline, link(logit) scheme(vg_palec) ///
ylabel(, nogrid angle(horizontal)) ///
xtitle("X (Units)", size(medium)) ytitle("Outcome Risk (%)", size(medium)) ///
graphregion(fcolor(white))
Read 7 tweets
Random thoughts about confidence intervals:

When we teach precision & accuracy, we often use images of a target, like these from the Wikipedia article: en.m.wikipedia.org/wiki/Accuracy_…

But I think this analogy leads to confusion about the interpretation of the confidence interval. 1/4
I’ve often heard students describe their confidence interval as if it were the fixed target, and the “true” value was the arrow that may or may not land on the target 95% of the time.... 2/4
But, the “true” value is the part they should view as fixed.

Instead, I propose explaining precision, accuracy, and confidence intervals with ring toss.

The “truth” is in a fixed place, and it’s the confidence interval ring that may or may not land where you want it to. 3/4
Read 4 tweets
More problems with logistic regression in the medical literature:
(PubMed listing, if you prefer: ncbi.nlm.nih.gov/pubmed/27756470)
Read 28 tweets

Related hashtags

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3.00/month or $30.00/year) and get exclusive features!

Become Premium

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!