My research ~process~
1. Excited about new project
2. New git repo
3. Data messier than I thought
4. Modeling harder than I thought
5. Guilt about other unfinished work
6. Wonder if any of it matters anyway given current state of world
7. Self-loathing, dread
8. Go to Step 1
Happy Halloween from us! The prior, likelihood and posterior (turns out getting a nice family photo with a baby that’s just learnt to crawl is impossible).
PhD: stressed because
- trying to do the best job I possibly can on one thing
- ignoring email from supervisor asking for updates
Assistant prof: stressed because
- trying to do a just okay job on 1254 things at once
- ignoring the 27 emails that have arrived since 10 min ago
This morning I ran a workshop on social media data for population researchers. We extracted Twitter data, and did some geocoding, mapping, and text analysis. All the materials and code are here for people who are interested:
Thinking about data collection: wow it would be great to collect info on all these variables! They would be super interesting to analyze!
Realities of data collection:
The toddler has a ‘Bayesian probability’ book so for the sake of a well-rounded education I made a frequentist version. The next page says “draw a ball from the urn, note its colour and return it. Repeat this an infinite number of times” so it should keep him busy for a while
Being in the same department as
@RohanAlexander
is a fun natural experiment. The same student emailed both of us with the same request.
They could have literally copy-pasted the greeting. And yet, Rohan gets "Dr" and I get "Ms"
This post runs through an example Bayesian workflow in
#rstats
, looking at the relationship between birth weight and gestational age.
There's code and examples on how to do prior and posterior predictive checks, and LOO-CV, using both Stan and brms.
This semester I taught stats to undergrad sociologists using the tidyverse in
#rstats
for the first time. Here's some reflections about what went well and what didn't:
I wrote a post about applied statistics, data, and answering hard questions. The takeaway is that when doing and teaching applied statistics, we need to emphasize engaging with data, and understanding data in context.
I'm recruiting a postdoc! Come and work on demographic methods, computational social science, and reproducibility. The position is for two years, with one year spent visiting the Lab of Digital and Computational Demography at
@MPIDRNews
. Details are here:
Someone found my wallet on the street in downtown Toronto. He saw my UofT ID card in it so googled me, emailed me and just now dropped it off at my door because the baby is napping. Canadians are amazing.
Sure I'll agree to do this recorded presentation for a webinar it's only 8 mins should be super quick to do
1395 takes, several hours later: "I'm Monica Alexander, an assistant professor in sadistics and sociology..."
Our
#rstats
package 'rcbayes' is now on CRAN. The package has functions to calculate, visualize and estimate Rogers-Castro migration schedules using
@mcmc_stan
.
A lot of my feed right now is different modeling takes for COVID-19 but one thing that sticks out is the demographers en masse yelling “YES BUT HAVE YOU CONSIDERED THE POPULATION AGE STRUCTURE” which is the most demography thing ever
It’s been a very long while between flights and I didn’t realize how much I missed the ‘coding while at an ordinary airport bar drinking a very ordinary beer’ vibe
Surely for the Monaco Grand Prix they should randomly assign drivers to cars, complete the race, repeat the process ten times and calculate the winner based on the best average time
#montecarloftw
Sociology at the University of Toronto is hiring a TT position in the area of the sociological study of science and technology! The department is full of wonderful, supportive, talented people, and the views from our new office are pretty solid
We have 4 TT jobs advertised at
@UofT_Sociology
this year! Areas are international migration, family, theory, and teaching stream. Toronto is great and the department is full of wonderful, talented people.
Links:
Found out today that a paper with
@LeontineAlkema
, which is older than both of my children, was accepted for publication so I am absolutely making cocktails
Today was my first time teaching in person in (almost exactly) two years. I forgot:
- what floor my office is on
- how projectors work
- how to make eye contact
- that people are different heights
Teaching statistics to undergraduate sociology students for the first time next semester and looking for a good intro textbook using
#rstats
...
@djnavarro
's 'Learning Statistics with R' is top of the list, but curious if people have other recommendations!
Acceptable greetings for academia-related emails:
If you know me: Monica
If you don’t know me: Dr. Alexander, Professor Alexander
What my inbox looks like:
Excited to see our paper with
@LeontineAlkema
published in
@ReadDemography
! We developed a cohort component projection framework that relies on limited data inputs, combined with demographic knowledge, to get projections and estimates of the components of population change
A new study by
@monjalexander
&
@LeontineAlkema
proposes a Bayesian CCP model to estimate subnational populations. Their age-specific net-migration model employs parameters that can be used in data-sparse settings to better understand drivers of change.
Happy Halloween from an (extra confused) confusion matrix!
Outcome of interest: person loves garbage trucks (yes/no)
Model predictions:
pre-schooler = yes (TP)
@RohanAlexander
= yes (FP)
toddler = no (FN)
me = no (TN)
Reviewing grad applications for another year and again in the rec letters the women are often described as "gentle and polite" and then men are often described as "wonderful and talented"
The 6 stages of a toddler having a cold:
1.Incubation
2.OMG DOES HE HAVE COVID? (No)
3.*google: How much Tylenol can toddlers have?*
4.Become at peace with not answering emails
5.*google: How many days can an adult survive without sleep?*
6.Oh now he’s fine but I feel like death
Hey
#poptwitter
! Inspired by recent discussions on here,
@v_dilego
,
@rmogimogi
and I have decided to start a formal demography working group. If you’re interested in being involved, sign up here:
There's currently an enormous wait time to talk to Passport Canada. But every minute they give you an update on the number of calls ahead of you. So I've turned it into a data collection exercise to use in my intro stats course next year
Very proud of
@RohanAlexander
for this remarkable book. Written in true Rohan style (weaving R examples with references to politics, philosophy, and novels), but also in a truly Rohan way: always listening and incorporating feedback from others. Pictured: behind the scenes edits
My book about introductory data science was published today. It is available online - - and you can buy a print copy: . I would appreciate your feedback and suggestions for improvements.
The next formal demography working group meeting is happening this Friday at 10am ET (Toronto time)! The formidable
@timriffe1
will be talking about decomposition methods and how to implement them.
Sign up at our website to be on the email list:
Maternal mortality vs data availability
The brighter the red, the higher the mortality; the brighter the blue, the better the data coverage.
Summary: much less data available for countries with a high mortality burden.
code to make map is here:
Super excited to join
@DemographicRes
as an Associate Editor! A great open access journal that accepts all types of submissions, including replications, research materials, and formal demographic relationships
📢We are excited to announce that the restructuring of our Editorial Board is complete and delighted to welcome all new members! See who’s who here ➡️ 🎉