Comments for Civil Statistician

Comment on Transitions by civilstat

civilstat — Thu, 11 Apr 2013 13:42:08 +0000

Thanks David! I do plan to keep up blogging what I learn at CMU. Cheers!

Comment on Transitions by david k waltz

david k waltz — Thu, 11 Apr 2013 09:12:40 +0000

Congratulations on several major life transitions! Blogs can be very useful ways to ensure new learning is committed to memory – kind of a virtual notebook that others can peek into. Good luck!

Comment on Transitions by civilstat

civilstat — Tue, 09 Apr 2013 16:45:49 +0000

Thanks for the well-wishes, Matt! And I’m glad you found value in the post (and of course in GoG)!

Comment on Transitions by Matt

Matt — Tue, 09 Apr 2013 16:37:43 +0000

Jerzy:

Congrats on your forthcoming marriage and entrance into the PhD program!

I just wanted to say thanks for all the awesome posts. It was your review of Grammar of Graphics that finally convinced me to buy the book (and I’m glad I did!).

Best of luck!

Comment on Making R graphics legible in presentation slides by Exporting nice plots in R » G-Forge

Exporting nice plots in R » G-Forge — Mon, 25 Feb 2013 21:19:28 +0000

[...] When increasing the resolution the labels automatically decrease and become unreadable [...]

Comment on Small Area Estimation resources by Data round up, February 6 | School of Data - Learn how to find, process, analyze and visualize data

Wed, 06 Feb 2013 12:44:36 +0000

[...] Estimation to improve the precision of your statistical estimates, check out this collection of reference materials and learning resources from Jerzy [...]

Comment on Small Area Estimation resources by civilstat

civilstat — Wed, 06 Feb 2013 02:16:00 +0000

Thanks, Harlan! Yep, many of the SAE approaches are mixed-effects models as you describe, but with a special focus on accounting for the survey weights and sampling variances explicitly.
In the examples I’ll post, you’ll see a model like , where i indexes the small areas, and the observed are already aggregated to the area level using sampling weights. So we usually also have a survey-design-based estimate of the area-level sampling variance that we often just treat as known. Then it’s just a matter of estimating the area-level random-effect variance and regression coefficients , then combining it all to get estimates of .
More details to follow!

Comment on Statistical Rules of Thumb, Gerald van Belle by Nova reinfeld-kirkman

Nova reinfeld-kirkman — Tue, 05 Feb 2013 23:55:56 +0000

just found you. nice work!

Comment on Small Area Estimation resources by Harlan

Harlan — Sat, 02 Feb 2013 14:14:14 +0000

Great resources! Looking forward to the longer treatment! I don’t do much survey research, but I have definitely done hierarchical modeling. I typically use mixed-effects/Bayesian approaches ala Gelman to “partially pool” subsets of the data towards the global trend. Are the SAE approaches you describe substantially different? How so?

Comment on audiolyzR: Data sonification with R by Jeff

Jeff — Mon, 14 Jan 2013 15:18:27 +0000

The similarity between this work and the fictional product “Anthem” in Douglas Adams’ book _Dirk Gently’s Holistic Detective Agency_ is pretty remarkable.