logobanner1
research label grad course label undergrad course label other labelbook linkarticles linkworking paper linkAdvanced Quantitative Political Methodology
	        linkmax likelihood linkvisualizing data linkpanel data linkPolitical Science as Social Science
	       LinkIntro to Soc Stat linkCase-Based Stat linkPolitical Economy Seminar linkSoftware linkData
	        link

Have a bug report?
Send to:

cadolph at uw dot edu


To install an R package from this site on a PC:

   

The package is now installed. Don't forget to load the library using R's library() function.

For a list of functions, e.g. for the tile package, use help(package=''tile'') at the R prompt.





R Packages

tile

Version 0.4.18    updated to work with R 4.3.2

       

A fully featured R graphics package built on the grid graphics environment.

tile plots data to one or more graphical displays in a tiled layout with automated (or customizable) titles, axes, and annotations, and can solve many graphics problems:

  • Make standard displays like scatterplots, lineplots, and dotplots
  • Create more experimental formats like ropeladders, nightplots, and riceplots
  • Summarize uncertainty in inferences from model
  • Avoid extrapolation from the original data underlying your model
  • Fully control titles, annotation, and layering of graphical elements
  • Build your own tiled graphics from primitives

Dependencies: Most examples in the tile package require the R package simcf, available on this page.

Demonstration scripts: “Tufte Without Tears” provides an overview of the package and several examples of tile in action. In addition, see the scripts below, and still more examples in the package help files.

Development:  tile is presently in development through this Github repository. If you would like to contribute, please contact Chris Adolph via the email at left.

simcf

Version 0.2.21    updated to work with R 4.3.2

       

Calculate counterfactual expected values, first differences, and relative risks, and their confidence intervals.

Currently supported models include linear regression, linear models with a lagged dependent variable, logit, probit, ordered logit and probit, multinomial logit, and log-linear models like Poisson and Negative Binomial regression.

simcf offers similar functionality to Zelig, but is more modular: users can use any function they like to do the initial statistical analysis, then use simcf to produce counterfactual simulations, and then plot the results of those simulations using any package they wish. This makes simcf particularly useful in conjunction with the tile package, and in particular its ropeladder and lineplot functions. The simcf + tile combination thus offers more control to advanced R users than Zelig, but at present supports a narrower range of models.

Dependencies: Often used in conjunction with the tile package provided above.

Demonstration scripts: See tile above.

Development:  simcf is presently in development through this Github repository. If you would like to contribute, please contact Chris Adolph via the email at left.

partialrank

Version 0.4 (alpha)

       

Bayesian estimation of partial rank regression models.

Partial rank data are observed up a known interval of ranks, called a tier. Information from known ranks can be used to estimate relationships among, and unknown values of, partially observed ranks. This package includes a Bayesian estimator for partial rank data, based on the rank likelihood, which accounts for the dependencies across different ranked observations. The package also contains censored regression and ordered probit approaches which do not account for these dependencies. Also included are prediction functions for each model, and random number generators for partial ranks.

Dependencies: R2WinBUGS, coda, lattice, MCMCpack (all available from CRAN)

Demonstration scripts: See help for partialRankBayes and run example(partialRankBayes) from the command line in R.

Typesetting Resources

caxetexFree

XeTeX style file, Version 0.12

A XeTeX stylesheet for scientific papers and books that uses only free OpenType fonts.

Now available on Overleaf! See the openly shared Overleaf project for caxetexFreeOL.

XeTeX is a LaTeX-like engine which allows for powerful OpenType font selection and thus truly general typesetting. XeTeX now comes bundled with MacTeX, so you may already have it installed on your system. Moreover, you probably won't need to adjust your tex file in order to compile using this stylesheet and XeTeX, other than to add \usepackage{caxetexFree} somewhere in the preamble.

This manual describes how to install and use the stylesheet; it also serves as a test document to see if your installation is working correctly (the underlying tex file is here, as well as the bib file and figures 1 and 2). To use caxetexFree, you will also need to download and install the following free fonts:


LibertineFreeAvailable here
Source Sans ProFreeAvailable from FontSquirrel
InconsolataFreeAvailable here; download both .ttf files.
Museo Slab 500FreeAvailable from FontSpring
Porson (Γρεεκ)FreeAvailable from the Greek Font Society

You will also need to compile your tex files with XeTeX rather than LaTeX. For Mac users who compile from Aquamacs, this is easy: from the Command menu, set your TeXing options to Use XeTeX engine. You should put caxetexFree.sty itself in the TeX path, or in the folder containing your tex file.

caxetexBook

XeTeX style file, Version 0.12

An alternate XeTeX stylesheet for scientific papers and books using commerical OpenType fonts.

This manual describes how to install and use the stylesheet; it also serves as a test document to see if your installation is working correctly (the underlying tex file is here). To use caxetexBook, you will also need to download and install the following fonts, some of which will cost you money:


Bembo Book Pro$Available from Fonts.com
Gill Sans$ Available here; comes free with Macs & Adobe CS
Archer$ Available from Hoefler & Frere-Jones
InconsolataFreeAvailable here; download both .ttf files.
Museo Slab 500FreeAvailable from FontSpring
Porson (Γρεεκ)FreeAvailable from the Greek Font Society

You will also need to compile your tex files with XeTeX rather than LaTeX. You should put caxetexBook.sty itself in the TeX path, or in the folder containing your tex file.

caletter

LCO file for the scrlttr2 class of KOMA-Script and XeTeX

My University of Washington letterhead, which UW students and faculty (or others) are welcome to adapt for their own use.

You will also need this template letter tex file and the UW logo (drawn from the University of Washington art pack). Here is a sample finished letter.

If you wish to use this letter format, you will need to become familiar with the scrlttr2 class for XeTeX. There is an extensive manual; focus on Chapter 6. This tutorial may also help. Your goal in editing the lco file is to change the personal information (address, name, title, etc.) to your own. You may also need to obtain/replace the typefaces loaded by caletter.lco; see the notes on caxetex above.

After adapting the lco file, run XeTeX on your letter tex file. For Mac users who compile from Aquamacs, this is easy: from the Command menu, set your TeXing options to Use XeTeX engine. Your lco file should be in the TeX path, or in the folder containing your tex file. You will need to supply your own scanned signature file, or omit that part of the code.

calatex

LaTeX style file

My old LaTeX stylesheet.

I now use the XeTeX style sheets above, but installing this stylesheet is easier – no nice fonts though!

Other Code Snippets

Escore

Gauss script

Functions for processing biographical data into time-specific, individual or organization-averaged scores of prior career experiences.

Details and documentation will be available when the R package implementation of Escore is completed.

Dependencies: None, but not tested in recent versions of Gauss.

avp

R script

Compute and plot actual-versus-predicted probability plots for binary choice models

Plotting the observed response against a regression model's predictions of that response can be a useful diagnostic tool and goodness of fit measure. Unfortunately, the binary nature of the response in logit (or probit or similar) models makes such plots essentially useless—unless we first bin the observations by ranges of the predicted outcome, and compute average predictions and average response rates within these bins. In the resulting plot, points near the 45° line, and especially points near the lower-left or upper-right corners of the plot, indicate a good fit. Patterns of points off the 45° line may provide hints to the source of misspecification causing a poor fit.

avp is an R function for producing this sort of plot, and can be loaded by downloading avp.r to the current working directory, and then running source("avp.r") from the R prompt. Documentation is currently unavailable, though most inputs should be self-explanatory. This function will eventually be folded into tile.

Demonstration scripts: This function is used in this demonstration script, which uses these data and produces three seperate outputs.




University of Washington link

CSSS Center for Statistics and the Social Sciences link

Designed by
Chris Adolph & Erika Steiskal

Copyright 2011–2022
Privacy · Terms of Use

Jefferson (2007-2011)