Trying to get html output because of problems with the projector

day-plan.md

@@ -5,10 +5,10 @@ author: |
   | ICPSR 2019 Session 2
   | Jake Bowers, Ben Hansen, Tom Leavitt
 bibliography:
- - refs.bib
+ - BIB/refs.bib
  - BIB/master.bib
  - BIB/misc.bib
- - ci.bib
+ - BIB/ci.bib
 fontsize: 10pt
 geometry: margin=1in
 graphics: yes
@@ -182,6 +182,7 @@ Balance tests and the Sequential Intersection Union Principle
  - Sensitivity analysis I (Rosenbaum Style)
 
 ### Readings on sensitivity analysis
+ - @rosenbaum2017observation, Chap 9
  - @rosenbaum10book, Chap 3
  - @hhh2010
  - @rosenbaumtwo

day10-Scores.Rmd

@@ -1,7 +1,7 @@
 ---
 title: Statistical Adjustment and Assessment of Adjustment in Observational Studies --- Matched Stratification for One and Multiple Variables.
 date: '`r format(Sys.Date(), "%B %d, %Y")`'
-author: ICPSR 2018 Session 2
+author: ICPSR 2019 Session 2
 bibliography:
  - BIB/abbrev-long.bib
  - BIB/refs.bib

day11-MVMatching.Rmd

@@ -1,17 +1,17 @@
 ---
 title: Matching on more than one covariate
 date: '`r format(Sys.Date(), "%B %d, %Y")`'
-author: ICPSR 2018 Session 2
+author: ICPSR 2019 Session 2
 bibliography:
- - refs.bib
+ - BIB/refs.bib
  - BIB/master.bib
  - BIB/misc.bib
 fontsize: 10pt
 geometry: margin=1in
 graphics: yes
-biblio-style: authoryear-comp
 biblatexoptions:
   - natbib=true
+biblio-style: authoryear-comp
 output:
   beamer_presentation:
     slide_level: 2
@@ -22,6 +22,16 @@ output:
     includes:
         in_header:
            - defs-all.sty
+    pandoc_args: [ "--csl", "chicago-author-date.csl" ]
+  revealjs::revealjs_presentation:
+    slide_level: 2
+    incremental: true
+    transition: none
+    fig_caption: true
+    self_contained: false
+    reveal_plugins: ["notes","search","zoom","chalkboard"]
+    pandoc_args: [ "--csl", "chicago-author-date.csl" ]
+    css: icpsr-revealjs.css
 ---
 
 <!-- Make this document using library(rmarkdown); render("day12.Rmd") -->
@@ -37,16 +47,18 @@ require(knitr)
 ## This plus size="\\scriptsize" from https://stackoverflow.com/questions/26372138/beamer-presentation-rstudio-change-font-size-for-chunk
 
 knitr::knit_hooks$set(mysize = function(before, options, envir) {
-  if (before)
-    return(options$size)
+			      if (before){
+				      return(options$size)
+			      } else {
+				      return("\\normalsize")}
 })
 
 knit_hooks$set(plotdefault = function(before, options, envir) {
     if (before) par(mar = c(3, 3, .1, .1),oma=rep(0,4),mgp=c(1.5,.5,0))
 })
 
 opts_chunk$set(
-  tidy=FALSE,     # display code as typed
+  tidy='styler',     # display code as typed
   echo=TRUE,
   results='markup',
   strip.white=TRUE,
@@ -86,16 +98,62 @@ library(chemometrics) ## for drawMahal
 library(mvtnorm)
 ```
 
+<!--- For HTML Only --->
+`r if (!knitr:::is_latex_output()) '
+$\\newcommand{\\Pdistsym}{P}$
+$\\newcommand{\\Pdistsymn}{P_n}$
+$\\newcommand{\\Qdistsym}{Q}$
+$\\newcommand{\\Qdistsymn}{Q_n}$
+$\\newcommand{\\Qdistsymni}{Q_{n_i}}$
+$\\newcommand{\\Qdistsymt}{Q[t]}$
+$\\newcommand{\\dQdP}{\\ensuremath{\\frac{dQ}{dP}}}$
+$\\newcommand{\\dQdPn}{\\ensuremath{\\frac{dQ_{n}}{dP_{n}}}}$
+$\\newcommand{\\EE}{\\ensuremath{\\mathbf{E}}}$
+$\\newcommand{\\EEp}{\\ensuremath{\\mathbf{E}_{P}}}$
+$\\newcommand{\\EEpn}{\\ensuremath{\\mathbf{E}_{P_{n}}}}$
+$\\newcommand{\\EEq}{\\ensuremath{\\mathbf{E}_{Q}}}$
+$\\newcommand{\\EEqn}{\\ensuremath{\\mathbf{E}_{Q_{n}}}}$
+$\\newcommand{\\EEqni}{\\ensuremath{\\mathbf{E}_{Q_{n[i]}}}}$
+$\\newcommand{\\EEqt}{\\ensuremath{\\mathbf{E}_{Q[t]}}}$
+$\\newcommand{\\PP}{\\ensuremath{\\mathbf{Pr}}}$
+$\\newcommand{\\PPp}{\\ensuremath{\\mathbf{Pr}_{P}}}$
+$\\newcommand{\\PPpn}{\\ensuremath{\\mathbf{Pr}_{P_{n}}}}$
+$\\newcommand{\\PPq}{\\ensuremath{\\mathbf{Pr}_{Q}}}$
+$\\newcommand{\\PPqn}{\\ensuremath{\\mathbf{Pr}_{Q_{n}}}}$
+$\\newcommand{\\PPqt}{\\ensuremath{\\mathbf{Pr}_{Q[t]}}}$
+$\\newcommand{\\var}{\\ensuremath{\\mathbf{V}}}$
+$\\newcommand{\\varp}{\\ensuremath{\\mathbf{V}_{P}}}$
+$\\newcommand{\\varpn}{\\ensuremath{\\mathbf{V}_{P_{n}}}}$
+$\\newcommand{\\varq}{\\ensuremath{\\mathbf{V}_{Q}}}$
+$\\newcommand{\\cov}{\\ensuremath{\\mathbf{Cov}}}$
+$\\newcommand{\\covp}{\\ensuremath{\\mathbf{Cov}_{P}}}$
+$\\newcommand{\\covpn}{\\ensuremath{\\mathbf{Cov}_{P_{n}}}}$
+$\\newcommand{\\covq}{\\ensuremath{\\mathbf{Cov}_{Q}}}$
+$\\newcommand{\\hatvar}{\\ensuremath{\\widehat{\\mathrm{Var}}}}$
+$\\newcommand{\\hatcov}{\\ensuremath{\\widehat{\\mathrm{Cov}}}}$
+$\\newcommand{\\sehat}{\\ensuremath{\\widehat{\\mathrm{se}}}}$
+$\\newcommand{\\combdiff}[1]{\\ensuremath{\\Delta_{{z}}[#1]}}$
+$\\newcommand{\\Combdiff}[1]{\\ensuremath{\\Delta_{{Z}}[#1]}}$
+$\\newcommand{\\psvec}{\\ensuremath{\\varphi}}$
+$\\newcommand{\\psvecgc}{\\ensuremath{\\tilde{\\varphi}}}$
+$\\newcommand{\\atob}[2]{\\ensuremath{#1\\!\\! :\\!\\! #2}}$
+$\\newcommand{\\stratA}{\\ensuremath{\\mathbf{S}}}$
+$\\newcommand{\\stratAnumstrat}{\\ensuremath{S}}$
+$\\newcommand{\\sAsi}{\\ensuremath{s}}$
+$\\newcommand{\\permsd}{\\ensuremath{\\sigma_{\\Pdistsym}}}$
+$\\newcommand{\\dZ}[1]{\\ensuremath{d_{Z}[{#1}]}}$
+$\\newcommand{\\tz}[1]{\\ensuremath{t_{{z}}[#1]}}$
+$\\newcommand{\\tZ}[1]{\\ensuremath{t_{{Z}}[#1]}} $
+'`
+
 ## Today
 
-\begin{enumerate}
-  \item Agenda:  Need to adjustment $\rightarrow$ "fair comparison" $\rightarrow$
+ 1. Agenda:  Need to adjustment $\rightarrow$ "fair comparison" $\rightarrow$
   stratification $\rightarrow$ Evaluation/Assessment of the stratification;
   How to do this with one variable. How to do this with more than one
   variable.
-\item Reading for tomorrow and next week: DOS 8--9, 13 and \cite[\S~9.5]{gelman2006dau}, and \cite{hans04} \cite{ho:etal:07}
-\item Questions arising from the reading or assignments or life?
-\end{enumerate}
+ 2. Reading for tomorrow and next week: DOS 8--9, 13 and \cite[\S~9.5]{gelman2006dau}, and \cite{hans:04} \cite{ho:etal:07}
+ 3. Questions arising from the reading or assignments or life?
 
 # But first, review:
 
@@ -107,32 +165,29 @@ library(mvtnorm)
    inference about causal effects).
    - The testing based approach (Fisher, Rosenbaum)
    - The estimation based approach (Neyman, Robins, many others)
-   - Skipped: The prediction based approach (Bayes, Rubin)
+   - Skipped: The prediction based approach (Bayes, Rubin); Estimation of Structural Models (Pearl)
  - Causal inference from simple randomized experiments.
  - Causal inference from randomized experiments with not-randomized doses
    (i.e. Instrumental Variables approaches to causal inference)
  - Briefly: How to know that our statistical procedures are doing what we want
    them to do: The operating characteristics of statistical procedures for
    esimation (bias,consistency) and testing (coverage / false positive error
-   rate, power). 
+   rate, power).
  - Making the case for adequate adjustment in observational studies
-   - The difficulties of making this case using the linear model
-   - The potential for making this case for one or more variables using
-     stratification.
-   - Using optimal, full matching technology to make and evaluate
-     stratifications.
+   - Using the linear model
+   - Using stratification  (via optimal, full matching) and the experimental standard
 
 ```{r echo=FALSE, cache=TRUE}
 load(url("http://jakebowers.org/Data/meddat.rda"))
-meddat<- mutate(meddat, 
+meddat<- mutate(meddat,
 		HomRate03=(HomCount2003/Pop2003)*1000,
                 HomRate08=(HomCount2008/Pop2008)*1000)
 row.names(meddat) <- meddat$nh
 ```
 #  Matching on Many Covariates: Using Mahalnobis Distance
 
 
-## Dimension reduction using the Mahalanobis Distance
+## Mahalanobis Distance
 
 The general idea: dimension reduction. When we convert many columns into one column we reduce the dimensions of the dataset (to one column).
 
@@ -143,7 +198,7 @@ plot(meddat$nhAboveHS,meddat$nhPopD,xlim=c(-.3,.6),ylim=c(50,700))
 ```
 
 
-## Dimension reduction using the Mahalanobis Distance
+## Mahalanobis Distance
 
 First, let's look at Euclidean distance: $\sqrt{ (x_1 - x_2)^2 + (y_1 - y_2)^2 }$
 
@@ -155,7 +210,7 @@ arrows(mean(X[,1]),mean(X[,2]),X["407",1],X["407",2])
 text(.4,200,label=round(dist(rbind(colMeans(X),X["407",])),2))
 ```
 
-## Dimension reduction using the Mahalanobis Distance
+## Mahalanobis Distance
 
 Problems with the Euclidean distance ($\sqrt{ (x_1 - x_2)^2 + (y_1 - y_2)^2
 }$): over/under-emphasis depending on scaling, ignores correlation.
@@ -174,12 +229,12 @@ tmp
 sqrt( (tmp[1,1] - tmp[2,1])^2 + (tmp[1,2]-tmp[2,2])^2 )
 ```
 
-## Dimension reduction using the Mahalanobis Distance
+## Mahalanobis Distance
 
-For the scaling problem, standardize the Euclidean distance:$\sqrt{ (x_1/sd(x_1) - x_2/sd(x_2))^2 + (y_1/sd(y_1) - y_2/sd(y_2))^2 }$. (Here also centering the variables, since we only care about distance.) 
+For the scaling problem, standardize the Euclidean distance:$\sqrt{ (x_1/sd(x_1) - x_2/sd(x_2))^2 + (y_1/sd(y_1) - y_2/sd(y_2))^2 }$. (Here also centering the variables, since we only care about distance.)
 
 ```{r echo=FALSE}
-Xsd <-scale(X) 
+Xsd <-scale(X)
 apply(Xsd,2,sd)
 zapsmall(apply(Xsd,2,mean))
 ```
@@ -193,7 +248,7 @@ text(2,-1.2,label=round(dist(rbind(colMeans(Xsd),Xsd["407",])),2))
 ```
 
 
-## Dimension reduction using the Mahalanobis distance
+## Mahalanobis distance
 
 
 
@@ -211,7 +266,7 @@ row.names(newX) <- row.names(X)
 cor(newX)
 ```
 
-```{r echo=FALSE, out.width=".5\\textwidth"}
+```{r echo=FALSE, out.width=".4\\textwidth"}
 mhnew <- mahalanobis(newX,center=colMeans(newX),cov=cov(newX))
 drawMahal(newX,center=colMeans(newX),covariance=cov(newX),
           quantile = c(0.975, 0.75, 0.5, 0.25))
@@ -228,9 +283,9 @@ text(newX[newpts,1]-.02,newX[newpts,2]-10,
 ## Dimension reduction using the Mahalanobis distance
 
 The contour lines show points with the same
-Mahalanobis distance and the numbers are Euclidean distance.
+Mahalanobis distance, the numbers are Euclidean distance. Notice that the point with Euclidean distance of 161 is farther than 250 in Mahalanobis terms.
 
-```{r echo=FALSE, out.width=".8\\textwidth"}
+```{r echo=FALSE, results="hide", out.width=".75\\textwidth"}
 mhnew <- mahalanobis(newX,center=colMeans(newX),cov=cov(newX))
 drawMahal(newX,center=colMeans(newX),covariance=cov(newX),
           quantile = c(0.975, 0.75, 0.5, 0.25))
@@ -434,7 +489,7 @@ plot(xb4)
 
 ## Calipers
 
-The optmatch package allows calipers (which disallow certain pairs from being matched).^[You can implement penalties by hand.] Here, for example, we disallow comparisons which differ by more than 2 standard deviations on the propensity score.
+The optmatch package allows calipers (which forbids certain pairs from being matched).^[You can implement penalties by hand.] Here, for example, we forbid comparisons which differ by more than 2 standard deviations on the propensity score.
 
 ```{r}
 quantile(as.vector(psdist),seq(0,1,.1))
@@ -443,7 +498,7 @@ as.matrix(psdistCal)[5:10,5:10]
 ```
 ## Calipers
 
-The optmatch package allows calipers (which disallow certain pairs from being matched).^[You can implement penalties by hand.] Here, for example, we disallow comparisons which differ by more than 2 standard deviations on the propensity score.
+The optmatch package allows calipers (which forbid certain pairs from being matched).^[You can implement penalties by hand.] Here, for example, we forbid comparisons which differ by more than 2 standard deviations on the propensity score.
 
 ```{r}
 fmCal1 <- fullmatch(psdist+caliper(psdist,2),data=meddat,tol=.00001)
@@ -551,8 +606,8 @@ strata or sets or selecting subsets.
  - The work on cardinality matching and fine balance
    <http://jrzubizarreta.com/>
    <https://cran.rstudio.com/web/packages/designmatch/> )
- - The work on speedier approximate full matching <http://fredriksavje.com/>
-   <https://github.com/fsavje/quickmatch>
+ - The work on speedier approximate full matching with more data  <http://fredriksavje.com/>
+   <https://github.com/fsavje/quickmatch>, <https://cran.r-project.org/web/packages/rcbalance/index.html>.
  - The work on using genetic algorithms to (1) find approximate strata
    with-replacement <http://sekhon.berkeley.edu/matching/> and (2) to find
    an approximation to a completely randomized study (i.e. best subset

day12-MatchingTools.Rmd

@@ -1,9 +1,9 @@
 ---
 title: Matching Tools and Approaches
 date: '`r format(Sys.Date(), "%B %d, %Y")`'
-author: ICPSR 2018 Session 2
+author: ICPSR 2019 Session 2
 bibliography:
- - refs.bib
+ - BIB/refs.bib
  - BIB/master.bib
  - BIB/misc.bib
 fontsize: 10pt
@@ -22,6 +22,15 @@ output:
     includes:
         in_header:
            - defs-all.sty
+    pandoc_args: [ "--csl", "chicago-author-date.csl" ]
+  revealjs::revealjs_presentation:
+    slide_level: 2
+    incremental: true
+    transition: none
+    fig_caption: true
+    self_contained: false
+    reveal_plugins: ["notes","search","zoom","chalkboard"]
+    pandoc_args: [ "--csl", "chicago-author-date.csl" ]
 ---
 
 <!-- Make this document using library(rmarkdown); render("day12.Rmd") -->
@@ -37,8 +46,10 @@ require(knitr)
 ## This plus size="\\scriptsize" from https://stackoverflow.com/questions/26372138/beamer-presentation-rstudio-change-font-size-for-chunk
 
 knitr::knit_hooks$set(mysize = function(before, options, envir) {
-			      if (before)
+			        if (before){
 				      return(options$size)
+			      } else {
+				      return("\\normalsize")}
 })
 
 knit_hooks$set(plotdefault = function(before, options, envir) {
@@ -88,11 +99,9 @@ library(mvtnorm)
 
 ## Today
 
-\begin{enumerate}
-  \item Agenda:
-  \item Reading for tomorrow and next week: DOS 8--9, 13 and \cite[\S~9.5]{gelman2006dau}, and \cite{hans:04} \cite{ho:etal:07}
-  \item Questions arising from the reading or assignments or life?
-\end{enumerate}
+  1. Agenda:
+  2. Reading for tomorrow and next week: DOS 8--9, 13 and \cite[\S~9.5]{gelman2006dau}, and \cite{hans:04} \cite{ho:etal:07}
+  3. Questions arising from the reading or assignments or life?
 
 # But first, review:
 

day12.Rmd

@@ -1,7 +1,7 @@
 ---
 title: The problems of covariance adjustment for bias; Simple stratification based approaches
 date: '`r format(Sys.Date(), "%B %d, %Y")`'
-author: ICPSR 2018 Session 2
+author: ICPSR 2019 Session 2
 bibliography:
  - refs.bib
  - BIB/master.bib

day13-MatchingInformation.Rmd

@@ -1,7 +1,7 @@
 ---
 title: Matching Tools and Information in a Block-Randomized Experiment
 date: '`r format(Sys.Date(), "%B %d, %Y")`'
-author: ICPSR 2018 Session 2
+author: ICPSR 2019 Session 2
 bibliography:
  - refs.bib
  - BIB/master.bib

day13.Rmd

@@ -1,7 +1,7 @@
 ---
 title: Matching on more than one covariate
 date: '`r format(Sys.Date(), "%B %d, %Y")`'
-author: ICPSR 2018 Session 2
+author: ICPSR 2019 Session 2
 bibliography:
  - refs.bib
  - BIB/master.bib

day14-MatchedEstimationTesting.Rmd

@@ -1,7 +1,7 @@
 ---
 title: Estimation and Testing for Stratified Matched Designs
 date: '`r format(Sys.Date(), "%B %d, %Y")`'
-author: ICPSR 2018 Session 2
+author: ICPSR 2019 Session 2
 bibliography:
  - refs.bib
  - BIB/master.bib

day14.Rmd

@@ -1,7 +1,7 @@
 ---
 title: Strategies for making matched designs
 date: '`r format(Sys.Date(), "%B %d, %Y")`'
-author: ICPSR 2018 Session 2
+author: ICPSR 2019 Session 2
 bibliography:
  - refs.bib
  - BIB/master.bib

day15.Rmd

@@ -1,7 +1,7 @@
 ---
 title: Information, Balance, Estimation, Testing
 date: '`r format(Sys.Date(), "%B %d, %Y")`'
-author: ICPSR 2018 Session 2
+author: ICPSR 2019 Session 2
 bibliography:
  - refs.bib
  - BIB/master.bib