Ian Cero

PhD in Clinical Psychology,
Masters in Probability and Statistics

[Ian Cero @IanCero]

We know that #suicide risk comes in clusters. Now we can use clustering of suicide-related posts to improve #SuicidePrevention.

Cero, I., & Witte, T. K. (2019). Assortativity of suicide-related posting on social media. American Psychologist. Preprint: https://t.co/FiErEPmQud

[Ian Cero @IanCero]

@righteousritard If you're interested in a light read that might help you appreciate math in a new way, you might try:

Strogatz, S. H. (2012). The Joy of x: a Guided Tour of Math, from one to Infinity. Houghton Mifflin Harcourt.

[Ian Cero @IanCero]

@rikcloesen @d08890 @lastpositivist And Gödel's incompleteness theorems?

[Ian Cero @IanCero]

@Yozarian22 @lastpositivist This gets my vote for top proposition to debate on the thread! Very interesting

[Ian Cero @IanCero]

@AGallyer @hardsci That's a little like saying nurses can protest by (a) not showing up to work or (b) never cashing their paychecks in an effort to mess with the payroll workflow.

Both would theoretically work, but the existence of the 2nd option doesn't help me understand why the 1st one is bad.

[Ian Cero @IanCero]

@AGallyer @hardsci I think you're clearly describing the morally optimal path: organizing around a plan

But it can also be true that non-organized review decliners still have an important effect. If wait times extended to, say, 5 years that would seem to put real pressure on a journal

[Ian Cero @IanCero]

@amybrausch @NeuroDrugDoc @lrkhazem @DocRobPhD23 @procAnna @kimvanorden Oh, sorry, I misunderstood the goal here. Kim and I are both researchers in a Psychiatry Department. We don't (as far as I know) have a mechanism to produce students who specialize in suicide without clinical training - though we DO take post docs without clinical training!

[Ian Cero @IanCero]

I know there are great packages for unit testing a package in R (e.g., testthat), but are there packages/frameworks for unit testing data wrangling and analysis?

#RStats

[Ian Cero @IanCero]

@IngramPsychLab Simon et al (2022) found their DBT module actually increased self-harm in a large RCT (n = 18,882), relative to TAU. JAMA, 327(7), 630-638.

It's likely many of those patients would have been better off with no treatment, despite the pre-trial plausbility of the DBT skills module

[Ian Cero @IanCero]

@NeuroDrugDoc I was just complaining to someone that this didn't exist and it would be a good public health measure if it did.

Looks like I have an apology to make...

[Ian Cero @IanCero]

@kearneymw Free online and very lucid. It's for python, but most of the conventions work across languages

https://t.co/cuyCs8jOOC

[Ian Cero @IanCero]

@EikoFried I feel the same way whenever I'm doing computations in parallel - or would that be with two trunks?

[Ian Cero @IanCero]

@CDelawalla @AGallyer @2KatherineMarie I agree neuro is over-funded, along with most basic psych work I see. But it's worth asking why we both think that

It's that such research has few historic successes in the real world, relative to time and money spent.

Sadly, that same standard applies to most applied psych too

[Ian Cero @IanCero]

@mitchprinstein @NIH @NewsHour @JudyWoodruff Moreover, I don't know of a collection of psychologists who - at the time - publically forecasted 60% uptake or less after a year.

Given our own work on the weakness of human forecasts, I dont think we can say with credibility that we as a field saw this state of affairs coming

[Ian Cero @IanCero]

@mitchprinstein @NIH @NewsHour @JudyWoodruff I admit I wouldn't have forecasted vaccine uptake would still be this low a year ago. Both US parties were still trying to claim credit for it, and vaccination rates were rapid for the first few months of 2021.

[Ian Cero @IanCero]

@AGallyer It may be that nothing works better than the control condition (e.g., nothing), but at least we would know that by the end and could make policy decisions accordingly.

Alternatively if some combo of uptake techniques did work well, we would have a good sense of what to try

[Ian Cero @IanCero]

@AGallyer I don't know what would have worked, but here's a study I would have liked to see funded and fast-tracked:

A large cluster-RCT with the top 2-3 most intuitive uptake techniques we have. Clusters would be semi-closed social units like schools, hospitals, or fire depts

[Ian Cero @IanCero]

@CDelawalla @AGallyer @2KatherineMarie I agree your work is important. But can you name a scalable intervention of the kind you're describing that has also had historic success in a large field experiment?

How is this different from behaviorists who made the same erronious forecast about their own work in the 50s?

[Ian Cero @IanCero]

@NeuroDrugDoc I'm not a Hayes fan either, which is why I regret having to take his side on this one. I worry that the last 18 months include a lot of empirical data in his favor - sadly, even outside the US too.

[Ian Cero @IanCero]

@NeuroDrugDoc So, there is (usually) good reason to think that limiting access to a popular method is helpful.

The second argument is that the substitutions that do occur are likely to go to a less lethal method (e.g., gun substituted with asphyxiation), so you still get a mortality reduction

[Ian Cero @IanCero]

@fourbeerspod @alexa_tullett

Quick followup on a great episode. If you're interested, excess suicides among dentists are likely a myth (Lange, 2012, Suicide rate in dental profession).

In contrast, veterinarians and physicians have a reliably high rate, even cross-nationally

[Ian Cero @IanCero]

@NeuroDrugDoc Sadly, we don't know what would happen in this specific case because we haven't tried it yet.

HOWEVER (!), in most cases where we've restricted access to lethal means before (e.g., coal gas, guns, pesticide), most people don't substitute methods (death certificates help confirm)