Political Science Computational Laboratory

Bayesian analysis of item-response theory (IRT) models, roll call analysis; computing highest density regions; maximum likelihood estimation of zero-inflated and hurdle models for count data; goodness-of-fit measures for GLMs; data sets used in writing and teaching at the Political Science Computational Laboratory; seats-votes curves.


News

1.5.2 * fixed bugs in calls to C routines in ideal.R

1.5.1 * AustralianElections includes 2010,2013,2016

  • presidentialElections includes 2016 and 2012 Alabama result
  • Manual updated to reflect changes

1.5.0 * improved truncated normal sampling (15-20% faster) * added use.voter option to ideal, implementing approach from Stephen Jessee's 2016 AJPS article. See ?ideal for details. * updated author contact information * fixed voteview URLs to reflect new site structure * fixed other broken URLs * CRAN compliance

1.4.9 * fixed bug in AIC correct in Vuong test

1.4.8 * package dependencies handled better * slight tweeks to ideal help * CRAN compliance * addressing memory leak issues revealed by valgrind in ideal's mallocs

1.4.7 * AIC and BIC in Vuong, better description and printing

1.4.6 * compliance with CRAN recss/reqs, vignettes sub-dir * better Imports/Depends/Suggests etc * dropped redundant "require" in examples etc * fixed long-standing buglet in Vuong test, AIC type df adjustment to test statistic * use format.pval for pvalues printing in Vuong output to screen

1.4.5 * better compliance with CRAN reccs/reqs, R 3.0.0 etc * drop leading zero in version number * compress PDFs from vignettes * lazy-loading, so um, lazy-load in readKH.r

1.04.4 * presidentialElections includes 2012

1.04.3 * bug in postProcess, not evaluating args in call of ideal object

1.04.2 * minor typo in dropRollCall.Rd

1.04.1 * fixed bug in non-English locales for hurdle/zeroinfl formula processing if second part of formula contained a period.

1.04 * fixed quite serious bug with storing item parameters * COPYING file deprecated (?), deleted from repos with r174

1.03.12 * minor bug in tracex with d>1 * deprecate showAll plotting option in tracex, change to "multi" (default=FALSE)

1.03.11 * small change to documentation for ca2006 (thanks Arthur Aguirre)

1.03.10 * pythag deprecated in Rmath.h, use system hypot instead (3/13/2011) * warnings about memory etc only come on with verbose=TRUE (req by Stephen Jessee)

1.03.9 * ideal: small change in partyLoyalty (thanks to Chris Hanretty) * ideal: reformat output of ideal to be 3-d arrays * ideal: change default prior precision for item parameters to .04 (used to be .01) * added UKHouseOfCommons data; Example 6.9 in BASS

1.03.8 * added an optional "at" argument to predprob() methods for count data so that the counts at which the probabilities are evaluated can be specified

1.03.7 * small bug in constrain.item (reported by Paul Johnson) * change normalization option in ideal to generate posterior means with mean 0, sd 1 * do normalization over all dimensions * typos in documentation for pseudo-R2 (thanks to Henrik P?rn)

1.03.6 * made gam dependency explicit * change linear.hypothesis to linearHypothesis

1.03.5 * added AustralianElectionPolling * tidy up Rd files for data sets (itemize -> describe) * use dQuote in Rd files (or not) * update AustralianElections with 2007 results

1.03.4 * added iraqVote * updated presidentialElections with 2008 results * added nj07 * added vote92

1.03.3 * improved offset handling in hurdle()/zeroinfl(): offsets in zero model are now allowed and can be different from count model. See ?hurdle/?zeroinfl for details. * included error message for negative binomial zero hurdle model with only an intercept as such models are not identified * fixed bug in predict() method for "hurdle" object when a Poisson zero hurdle model is employed * fixed bug in formula processing of hurdle()/zeroinfl(): formulas of type "y ~ . | . - x1 - x2" were not processed correctly. * for weighted hurdle/zeroinfl models the number of observations is now the effective number, i.e., with weights > 0

1.03.2 * fixing .Last.lib problems * turned off MDA option in ideal by default, seems broken, needs work

1.03a * make startvalues in ideal accept names other than xstart (anything unique that starts with x will do)

1.03 * bug in ideal start value argument processing thanks to Bjorn Hoyland

1.02 * added politicalInformation data * added MDA option to ideal

1.01 * added EfronMorris data * added RockTheVote data

1.00 * updated CITATION and docs to version 1.00 accompanying JSS paper

0.97 * minor changes to print and summary methods for rollcall and ideal objects * bugs in ideal (discovered via odd combination of store.item and file) * dropRollCallObject makes prettier output, works a little more sanely * s109 is now updated to its final status, example points to voteview.com * s110 in examples for readKH * adding component named "dropInfo" to rollcall objects produced by dropRollCall, per request of Keith Poole (this stores information about which legislators/votes are retained by dropRollCall) * added presidentialElections data from Jackman book manuscript * bugs in igammaHDR fixed, some typos too

0.96 * Final changes for JSS paper (before release 1.0 accompanying JSS paper). * Changed the default formula specification for zero-inflated models. y ~ x now means y ~ x | x and not y ~ x | 1 anymore. Thus, the same default formula processing is applied to zero-inflated and hurdle models. * Error in normalizeIdeal, found by Jeff Lewis

0.95 * fixed fitted() and predict() methods for "hurdle" and "zeroinfl" for the case when an offset is specified (this had been erroneously ignored previously)

0.94 * added analytical gradients for hurdle() and zeroinfl() (speed-up of factor 5-10) * new package "sandwich" (2.1-0) provides estfun() and bread() methods for "hurdle" and "zeroinfl" objects, thus sandwich() and vcovOPG() covariances are available * extended/improved countreg vignette * fixed Pearson residuals to use correct zero-augmented variance * extended predict() methods: type "count" and "zero" * added extractAIC() methods

0.93 * improving matching legislator and vote names in constrain functions, in response to sub-optimal behavior reported by Keith Poole

0.92 * minor errors in various functions spotted with stricter code checking in R-devel 2.6 * unescaped "%" in documentation for s109 * need negative 1 for negative intercept in another part of predict.ideal (for the case where the researcher did not supply burn-in; 05/15/2007).

0.91 * added Monte Carlo simulation of pi (simpi)

0.90 * version bump as requested by Achim * adding pseudo r-squared stuff * adding admit data for ordered probit example * adding logLik.polr * adding hitmiss method

0.76 * offset and weights arguments have been added to zeroinfl() and hurdle() * hurdletest() is a new wrapper for linear.hypothesis() from the "car" package to test for the presence of hurdles. * A vignette about count data regression in R has been added (which in particular explains the ideas behind zeroinfl() and hurdle()). * correcting/improving documentation of ideal re identification and the normalize option * added absentee ballot data from Orley Ashenfelter, used in BASS * John Fox noted an error in odTest (wrong p-value); fixed.

0.75 * version bump, requested by CRAN maintainers * minor typo in help for predict.ideal * fixed bugs with interactive() in ideal.r * trouble-shooting predict.ideal as per request of Boris Shor * bug (?) in predict.ideal; after 0.73 change to negative intercept, need change in predict.ideal.r to match * in zeroinfl(): binomial(link = link) is deprecated, use binomial(link = linkstr) instead

0.74 * deprecating meanzero option in ideal, replace with normalize * new options for specifying start values in ideal (eigen, random, or user-supplied list); this fixes a bug reported by Boris Shor [email protected] when working with large rollcall objects * removed bad escape "," in readKH reported by Kurt Hornick [email protected] * dummy entry to force svn commit/update (testing new svn server)

0.73 * added seatsVote class and ca2006 data * cleaned up startvalues processing in ideal * improved priors in ideal * added verbose option to ideal * less console output from ideal.c * some changes to documentation, e.g., postProcess gets extra references * sort output by posterior means of ideal points in summary.ideal * improved tracex * fixed bug in dropRollCall (was dropping legis.data attributes) * fixed bug in partyLoyalty, called by summary.rollcall(...,verbose=TRUE) * finally changed ideal model to have a negative intercept; required changes to updatex.c, updatey.c, and xreg.c

0.72 * fixed error in bioChemists data found by Bettina Gr?n [email protected], variable kids5 was off by 1 unit, now runs from min of zero (no kids).

0.71 * fixed bug in betaHPD discovered by John Bullock

0.70 * completely rewritten version of hurdle() and zeroinfl(): - new formula interface of type y ~ x | z where y ~ x specifies the count model and z the inflation/hurdle regressors. - re-structured returned value, is now more similar to "glm" objects - extended/enhanced extractor functions

0.62 * plot.ideal.1d: better left plot margin, based on max length of legis.name * plot.ideal.2d: inconsistent testing of presence of beta in ideal object when overlaying cutting planes * plot.ideal.Rd: more examples (but in \dontrun) * tracex: bug for 2d ideal objects * tracex: 2d, make legend lines heavier for showAll * tracex: for R >= 2.4, change par() to par(no.readonly=TRUE) * fixed typo in plot.ideal.Rd

0.61 * fixed bug in summary.ideal found by Keith Poole (8/8/06) * documentation of ideal section on Identification changed to reflect presence of postProcess function * added summary.rollcall into NAMESPACE per request of Jeff Lewis * no nsl function on Windows, changed readKH to simply call readLines with a try-error wrapper per suggestion of XP-user Jeff Lewis * processing of dropList improved, per suggestion of Jeff Lewis * cosmetic changes to print.summary.rollcall * fixed errors caused by removing sysdata.rda; now utility data sets such as stateinfo and partyinfo have to loaded via data() in functions inside readKH subfunctions; again, Jeff Lewis reported error testing readKH function.

0.60 * added postProcess * added jitter to plot.predict.ideal * changed "start" to "burnin" in many function that handle ideal objects * changed checkStart to checkBurnIn in idealHelper.r, returns "keep" (logical vector) * improved documentation for plot1d * dropped plot1d and plot2d from export in NAMESPACE * consolidated some help files, making fewer entries in package index etc * checked package with codetools(), cleaned up some redundancies, overwriting of function args inside of functions etc * minor stylistic changes to titles and descriptions in help files

0.59 * fixed compiler warning in dtnorm regarding return value * fixed compiler warning in IDEAL regarding double printed as int * minor documentation fix * removed unneeded directories

0.58 * documentaton fixes and cleanups, prompted by Kurt Hornik's check of 0.57 at CRAN * fixed errors in rollcall documentation

0.57 * added dtl file support to readKH * extensive reliance on match.call(), rather than carrying around copies of objects, many changes to numerous functions * dropList introduced, many functions changed; see dropRollCall and dropUnanimous (rollcall objects) * added extractRollCallObject (for ideal objects) * added computeMargins, for rollcall objects * added convertCodes, operates on codes component of roll call object * tracex handles 2d traces gracefully * added plot.predict.ideal * prior precisions in constrain.legis and contstrain.item are .01 for unconstrained parameters * added s109, dropped s102 as demonstration data, ships with package * added vectorRepresentation (rollcall object)

0.56 * changed rollcall class to have legis.data and vote.data data frames * amended plotting functions appropriately * rollcall objects have party loyalty scores * rollcall objects have lopsided data; summary.rollcall/print.summary.rollcall will display * ideal saves some barebones summary statistics (posterior means) * summary.ideal computes standard deviations * summary.ideal/print.summary.ideal displays information on bills that fail to discriminate * plot1d (plot method for 1d ideal objects) has option for all legislator names * plot2d (plot method for 2d ideal objects) has option to overlay estimated cutting lines * added readKH function and supporting data objects in R/sysdata.rda (state.info and party) * readKH has extensive checks and debugging of reads from web (thanks to Kurt Kornik [email protected]) * minor bug fixes in documentation and examples * made ideal examples shorter * add coda to list of required packages * error in negative binomial hurdle model, added theta to coefficients dimnames * added TODO file to top-level directory of package * fixed bug in print.zeroinfl (thanks to Bettina Gruen [email protected]) * bugs in zeroinfl (Bettina Gruen [email protected]) * corrected spelling of Ginsburg in sc9497 (Supreme Court sample data)

0.55 * added ideal point estimation (Alex Tahk) * added gamma functions for Jackman BASS book

0.54 * fixed errors in documentation for summary.zeroinfl

0.53 * edited documentation for odTest

0.52 * changed print.matrix references to print, since print.matrix deprecated in R > 2.1 * added ntable from Jim Fearon * added betaHPD

0.51 * fixed error in print.summary.zeroinfl (thanks to Dave Atkins; [email protected])

0.5 * initial relase to CRAN

Reference manual

It appears you don't have a PDF plugin for this browser. You can click here to download the reference manual.