Thread by @EmilyHGriffith1

Full room on a Thursday morning to learn about making an impact with #statistics in the era of #DataScience at #JSM2019

Data Science = w*Statistics + (1-w)*Computer Science
What a great definition!! #JSM2019

Sparsity and complexity are a defining aspect of big data. You need knowledge of the problem you’re solving, and that needs to be baked into our models. #JSM2019

Great rule from @hadleywickham : “If your first plot doesn’t reveal a data quality problem, there’s a data quality problem you have not yet revealed.” #JSM2019 #truth

Xiao-Li Meng says: Where is the right resolution for inference on an individual? We have so much data that we end up with no data! #JSM2019

It’s nice to hear multiple panelists recognize the value of using #statistics to solve applied problems even when the math isn’t fancy. #JSM2019 #ifeelseen

#DataScience is too broad to be squeezed into one academic department! #JSM2019

Xiao-Li Meng compares #DataScience training to medical training: we need specialists! #JSM2019

People like to reinvent the wheel, but we don’t need to. Look to disciplines that already do the work: OR, IEEE, library and information sciences. #JSM2019

Successes of the past 20 years? Jeffrey Leek says individuals have been doing amazing work, but institutions need to improve. #JSM2019

Tian Zheng says a major success is teaching the importance of exploratory data analysis. #JSM2019

.@hadleywickham suggests we overemphasize “statistical monogamy” and should focus on “safe stats.” 😂 #JSM2019

Where have we failed?? P-values. Where have we succeeded? Effect sizes. #JSM2019

Communication is absolutely crucial— this is something we don’t necessarily do well (as a field). #JSM2019

Statisticians need to show up! Meet people where they are. We cannot thing that if we’re good enough, people will find us. #JSM2019

I could listen to Xiao-Li Meng tell stories all day! He’s telling a great story about improving someone’s methods and ending up with a wider interval estimate. #JSM2019

How can we explain the value we add when making an estimate “worse”? #JSM2019

John Quakenbush is discussing the importance of working backwards— statistics is a creative part of working through an experiment. It’s not black magic that produces a p-value. #JSM2019

.@hadleywickham asks how can we make it as easy as possible for people to do the right thing? #potholeofsuccess #JSM2019

Project-based learning on real problems can help students learn to be confident in their usefulness! Thanks Tian Zheng! #JSM2019

Be a teammate, nor a referee! Everyone hates the referee— we should fail and succeed with our teams. — Jeffrey Leek #JSM2019

Don’t wait to be invited to the #DataScience party— start a conversation and take that first step. — Tian Zheng #JSM2019

We need to recognize that we live in a society that doesn’t trust science, data, or evidence-based decision making. #JSM2019

We need to lead people to understand what data and evidence means. We can have a big impact by sharing these tools. — John Quakenbush #JSM2019 #truthiness

Statistical thinking is a concept being mentioned again and again. It’s so interesting to hear how all the panelists (and talks from the whole conference) really are aligned on these foundational issues. #JSM2019

New question: how can we change our rewards system (particularly in academia) to recognize a broad array of faculty contributions? #JSM2019

John Quakenbush emphasizes the importance of translating collaborative work into a language more traditional senior folks can understand. #JSM2019

Xiao-Li Meng gives a dean’s perspective: we can’t change our field without changing other fields. We need to create more awards to give to people. That can help show our value! #JSM2019

We need to emphasize the intellectual challenges of what we do! (Math does this well.) we also need metrics! #JSM2019

Senior people have an important role to play here! They’re the ones with the power in this situation. #JSM2019

Senior faculty have an obligation to hire people who don’t look like them and don’t do the same thing. That’s a short -term solution. #JSM2019

It’s not that hard to come up with new metrics to assess each other! We are statisticians and can figure this out. — Jeffrey Leek #JSM2019

The long-term solution is to burn it all down and start over! #statisticalrevolution #JSM2019

Tian Zheng wants us to move away from looking for the “perfect” incoming student in the admissions process. #JSM2019

.@hadleywickham asks what skills do we need to be successful? Email management, communication, kindness, writing! #JSM2019

Be forward thinking and create a pathway for our students to let them become successful quantitative scientists. — John Q. #JSM2019

Xiao-Li Meng’s discussing the importance of leadership training (and he expressed interest in taking the course on email!). #JSM2019

Admissions should include more than just mathematical ability. It’s a solvable problem! #JSM2019

Go solve problems you think are interesting!

A lot of people need help— remember that! #JSM2019

This panel has been SO encouraging, interesting, uplifting, and funny! #JSM2019

First audience question is about moving past talking to concrete next steps. The panel seems optimistic. Think globally, act locally. #JSM2019

Question about researching in industry vs academia. Panel suggests finding the funding to do what you want, then go there! #JSM2019

There was a plea for at least some cool people to stay in academia! 😂😂😂 #JSM2019

@ReginaNuzzo asks about data scientists collaborating with journalists. A: There should be synergism between the two. We could learn a lot from each other! #JSM2019

Data journalism is a party statisticians should invite themselves to. Journalists tend to be chatty and outgoing (maybe unlike us?). #JSM2019

Q: is it important to teach students to fail? And if so, how do we give them safe ways to fail? #JSM2019

A: Yes! Failing isn’t failing, failing is part of research. Again, project-based learning has a big role here. Think about the Suzuki method for data science! #JSM2019

We should all fail together and then we can celebrate our eventual improvement! #ilovethis #JSM2019

How do you fail as fast as possible? That will help you iterate your way to success. — .@hadleywickham #JSM2019

Q: How can statistics pick up the pace? A: 1. Publish in archive. 2. It’s a complicated question— there’s a downside to speed. #JSM2019

Don’t rely solely on journals to advertise your work! #JSM2019

Thank you to the panel!! #JSM2019