Lightnews — Scholar-powered news

shira mitchell @shiraamitchell.bsky.social · 1d

CC: @tslumley.bsky.social

shira mitchell @shiraamitchell.bsky.social · 1d

blog post: struggles with equivalent weights

you've done MRP.

someone asks you for survey weights.

how to get them ?

1 1 3

shira mitchell @shiraamitchell.bsky.social · 7d

CC @fontikar.bsky.social @njtierney.bsky.social 🥾

shira mitchell @shiraamitchell.bsky.social · 7d

blog post: beyond balancing

in midterms, voters tend to support the out party for balance

do polls still help predict midterms ? yes

1 1 1

Reposted by shira mitchell

Gothamist @gothamist.com · 12d

The ultimate New York City lover’s treasure hunt is back Friday, Oct. 17 through Sunday, Oct. 19: bit.ly/3IHnHwn

2 4

shira mitchell @shiraamitchell.bsky.social · 14d

highly recommend reading @soodoku.bsky.social's post (even if it is about elephants and not bears)

www.gojiberries.io/lessons-from...

The Elephant in the Sampling Frame: Lessons from Basu's Parable

In Basu's circus story (PDF), the owner wants to estimate the weight of fifty elephants and suggests weighing Sambo, who looked “average” three years ago, and reporting $50y$. A statistician counters ...

www.gojiberries.io

1 1

shira mitchell @shiraamitchell.bsky.social · 15d

it was for me too !

1

shira mitchell @shiraamitchell.bsky.social · 15d

blog post: Fat Bear Week

Basu's Bears is a lesson in:

1) using auxiliary information (pre-salmon-feasting weights)

2) how bad an unbiased estimator can be

statmodeling.stat.columbia.edu/2025/09/23/s...

2 3 2

shira mitchell @shiraamitchell.bsky.social · 22d

CC: @shirokuriwaki.bsky.social @rnishimura.bsky.social

shira mitchell @shiraamitchell.bsky.social · 22d

blog post: random sampling is not leaving

we turned to response instrument Z because random sampling is "dead"

but does this method still rely on starting with random sampling ?

1 2 3

shira mitchell @shiraamitchell.bsky.social · 22d

blog post: random sampling is not leaving

we turned to response instrument Z because random sampling is "dead"

but does this method still rely on starting with random sampling ?

1 1

shira mitchell @shiraamitchell.bsky.social · 27d

that makes sense !

Z0 = treatment at time 0
Z1 = treatment at time 1
Y1 = outcome at time 1
Y2 = outcome at time 2

teammate 1 is regressing Y1 on Z0
teammate 2 is regressing Y2 on Z0, Z1, Y1

teammate 2 wants Y1 not to be imputed with Y2, right ?
teammate 1 wants Y1 to be imputed with Y2, right ?

shira mitchell @shiraamitchell.bsky.social · 27d

thanks, Lucy ! What if each teammate is running a different analysis with a different "Y" ? my "Y" might be your "X" ? would we have a different imputation procedure for each of us so that each person's "Y" is no used in imputing their "X" ?

2

shira mitchell @shiraamitchell.bsky.social · 29d

blog post on imputation (again):

we want E[Y|X] but X can be missing

@lucystats.bsky.social @sarahlotspeich.bsky.social @glenmartin.bsky.social @maartenvsmeden.bsky.social et al. say:

random imputation should use Y
deterministic imputation shouldn't

statmodeling.stat.columbia.edu/2025/09/09/s...

2 1 2

shira mitchell @shiraamitchell.bsky.social · 29d

I learn from every interaction with Maria. I attended a talk of hers yesterday. I highly recommend reading her work.

Maria Gargiulo @thegargiulian.bsky.social · Jun 4

🧮 A new preprint of my first dissertation chapter "From disparate lists to population estimates: A multiple systems estimation workflow for mortality analysis in conflict settings” is now online.

doi.org/10.31235/osf...

2

shira mitchell @shiraamitchell.bsky.social · Sep 3

blog post: connections between survey statistics and experimental design.

split-plot designs are analogous to cluster sampling.

blocking is analogous to stratification.

featuring an experiment by Arjun Potter and colleagues at NM-AIST !

7 16

shira mitchell @shiraamitchell.bsky.social · Aug 28

CC @davidbroska.bsky.social @austin-van-loon.bsky.social

1 1

shira mitchell @shiraamitchell.bsky.social · Aug 27

💧🥞🐢🐶

1

shira mitchell @shiraamitchell.bsky.social · Aug 27

blog post: Thomas Lumley writes about Interviewing your Laptop

what are the problems with using LLMs as survey respondents ?

how are these similar to problems with poststratification ?

CC @tslumley.bsky.social

1 1 4

Reposted by shira mitchell

soodoku @soodoku.bsky.social · Aug 25

Now with 'nearly isotonic' ...

github.com/finite-sampl...

cc @shiraamitchell.bsky.social

GitHub - finite-sample/rank_preserving_calibration: Rank preserving calibration of multiclass prob.

Rank preserving calibration of multiclass prob. Contribute to finite-sample/rank_preserving_calibration development by creating an account on GitHub.

github.com

2 1

shira mitchell @shiraamitchell.bsky.social · Aug 24

We are hiring !

Data Scientist – Analytics: job-boards.greenhouse.io/blueroserese...

and

Research Analyst: job-boards.greenhouse.io/blueroserese...

Data Scientist - Analytics (Senior and Junior)

Remote

job-boards.greenhouse.io

1 2

shira mitchell @shiraamitchell.bsky.social · Aug 23

‘it can be used as a drawing medium’ !! (Wikipedia)

1 1

Reposted by shira mitchell

Andrew Gelman et al. @statmodeling.bsky.social · Aug 19

Survey Statistics: answers from the BLS
statmodeling.stat.columbia.edu/2025/08/19/s...

Survey Statistics: answers from the BLS | Statistical Modeling, Causal Inference, and Social Science

statmodeling.stat.columbia.edu

2 7

shira mitchell @shiraamitchell.bsky.social · Aug 19

excited to soon get these references !

1

shira mitchell @shiraamitchell.bsky.social · Aug 19

blog post: answers from the BLS

2 weeks ago we learned about the CES employer survey that produces the jobs count.

we asked: why use employment size in stratification but not nonresponse adjustment ?

BLS responded !

statmodeling.stat.columbia.edu/2025/08/19/s...

1 2