#LyX 1.3 created this file. For more info see http://www.lyx.org/
\lyxformat 221
\textclass article
\language english
\inputencoding auto
\fontscheme times
\graphics default
\paperfontsize 12
\spacing single 
\papersize Default
\paperpackage a4
\use_geometry 1
\use_amsmath 0
\use_natbib 0
\use_numerical_citations 0
\paperorientation portrait
\leftmargin 1in
\topmargin 1in
\rightmargin 1in
\bottommargin 1in
\secnumdepth 3
\tocdepth 3
\paragraph_separation indent
\defskip medskip
\quotes_language english
\quotes_times 2
\papercolumns 1
\papersides 1
\paperpagestyle default

\layout Standard

Paul Johnson
\layout Standard

POLS 707 Research Methods II
\layout Standard
\added_space_top medskip 
Top 13 Things worth knowing about R.
\layout Enumerate

S books are R books.
\newline 

\newline 
S is a statistical language developed at Bell Labs (R.A.
 Becker, J.M.
 Chambers, A.R.
 Wilks, 
\emph on 
The New S Language.

\emph default 
 Pacific Grove, CA: Wadsworth & Brooks/Cole, 1988) and a computer program
 that speaks S, called 
\begin_inset Quotes eld
\end_inset 

S
\begin_inset Quotes erd
\end_inset 

 (or, more recently, 
\begin_inset Quotes eld
\end_inset 

S+
\begin_inset Quotes erd
\end_inset 

) is for sale from Mathsoft, Inc.
 
\newline 

\newline 
R is a statistical language very much like S, better in some ways, and it
 is available for free and in open source code.
 R is created/developed by a community of statisticians and programmers
 who truly desire to make good tools available for research and escape the
 binds imposed by commercialization of software.
\newline 

\newline 
Books about S generally will also be applicable to R.
 Here are some of the really useful ones of which I have copies.
\newline 

\newline 
W.N.
 Venables and B.D.
 Ripley, 
\emph on 
Modern Applied Statistics with S
\emph default 
, 4th edition.
 Brian Ripley is an amazing guy.
 He must work 20 hours per day.
 He is one of the primary developers behind the R movement and he reads
 the r-help email list.
 If you ever touch on some problem deep down in the guts of R, he may be
 the only one who will really know the answer.
 
\newline 

\newline 
Peter Dalgaard, 
\begin_inset Quotes eld
\end_inset 

Introductory Statistics with R
\begin_inset Quotes erd
\end_inset 

.
 Peter Daalgard is very active in r-help and also he is a very pleasant
 person and a good writer.
 This book is very well done.
 Not so advanced as V&R, but much more understandable to your average user.
 This shows how to complete many 
\begin_inset Quotes eld
\end_inset 

garden variety
\begin_inset Quotes erd
\end_inset 

 tasks.
\newline 

\newline 
Frank E.
 Harrell, Jr.
 
\emph on 
Regression Modeling Strategies: With Applications to Linear Models, Logistic
 Regression, and Survival Analysis.
 
\emph default 
This book has great practical information about how to do projects.
 He has complete code and working examples.
 Especially important if you want to use Harrell's advanced R packages,
 Hmisc and Design.
 Frank Harrell is (like me) a 
\begin_inset Quotes eld
\end_inset 

refugee
\begin_inset Quotes erd
\end_inset 

 from SAS and S+.
 He was the one who wrote the first 
\begin_inset Quotes eld
\end_inset 

Proc Logistic
\begin_inset Quotes erd
\end_inset 

 that was available in SAS in the early 1980s.
 In the 1990s he was working in S and S+.
 
\newline 

\newline 
The S-Plus stat manuals are available for free, in pdf format!
\newline 

\begin_inset LatexCommand \url[S-Plus site: ]{http://www.insightful.com/support/documentation.asp?DID=3}

\end_inset 


\newline 
You especially want the Guide to Statistics, Parts I and II.
 These have simpler explanations of many models than you will find in other
 places.
 The syntax for usage is not exactly the same as R, but they are worth downloadi
ng and saving for reference.
\layout Enumerate

R has built-in help.
\begin_deeper 
\layout Enumerate

Procedure specific help.
 If you want to know more about the 
\begin_inset Quotes eld
\end_inset 

hist
\begin_inset Quotes erd
\end_inset 

 procedure, type
\newline 
> help(hist)
\newline 
That 
\begin_inset Quotes eld
\end_inset 

help
\begin_inset Quotes erd
\end_inset 

 method gets used so often that they created a short-cut for it:
\newline 
> ?hist
\newline 
and after you look that over, you notice in the bottom of the help page
 it has examples, and if you want to see the examples, do
\newline 
> example(hist)
\newline 
If the examples all whir by too quickly, use this command to cause R to
 ask you to hit the return key between pictures:
\newline 
> par(ask=T)
\newline 
and then type your example(hist) command again.
\layout Enumerate

Get help inside your web browser.
 Type 
\newline 
> help.start()
\newline 
and watch what happens.
 Your web browser should show you a top level view, which includes links
 to the 
\begin_inset Quotes eld
\end_inset 

Introduction to R
\begin_inset Quotes erd
\end_inset 

 book as well as the manual on data import and export.
\newline 

\newline 
You will see also there is a FAQ for R in the online docs.
 The R FAQ is not a list of precise details about how to do certain things,
 but rather a higher level explanation of the R language and the various
 platforms on which it is used.
\layout Enumerate

Type
\newline 
> help.search(
\begin_inset Quotes eld
\end_inset 

hist
\begin_inset Quotes erd
\end_inset 

) 
\newline 
and watch what happens.
 This returns a giant list of methods, most of which you don't really want.
 But some you do.
 Alternative versions of histogram methods appear.
 For example, Brian Ripley does not like the default 
\begin_inset Quotes eld
\end_inset 

hist
\begin_inset Quotes erd
\end_inset 

 method, so his package MASS includes 
\begin_inset Quotes eld
\end_inset 

truehist
\begin_inset Quotes erd
\end_inset 

.
 The MASS package's truehist is more readily configurable.
\end_deeper 
\layout Enumerate
\added_space_bottom medskip 
Every R command needs parentheses!
\newline 
Even when you want out of R, you can't just type quit.
\newline 
You type
\newline 
>quit()
\newline 
or, for short,
\newline 
> q()
\newline 
If you type 
\newline 
> q
\newline 
R thinks you want to print out the contents of something named 
\begin_inset Quotes eld
\end_inset 

q
\begin_inset Quotes erd
\end_inset 

, but it can't find such a thing, and it tells you so.
\layout Enumerate


\begin_inset Quotes eld
\end_inset 

Equal to
\begin_inset Quotes erd
\end_inset 

 means a whole different thing than you expect.
\newline 
If you create a new variable, do not use the equal sign (
\begin_inset Formula $=$
\end_inset 

).
 Instead, you must use the symbol 
\begin_inset Formula $<-$
\end_inset 

to 
\begin_inset Quotes eld
\end_inset 

assign
\begin_inset Quotes erd
\end_inset 

 a value.
 So, for example, you can set a constant named 
\begin_inset Quotes eld
\end_inset 

b
\begin_inset Quotes erd
\end_inset 

:
\newline 
> b 
\begin_inset Formula $<-$
\end_inset 

3.3
\newline 
or define a vector named 
\begin_inset Quotes eld
\end_inset 

b
\begin_inset Quotes erd
\end_inset 

:
\newline 
> b 
\begin_inset Formula $<-$
\end_inset 

c(3,2,1,4,1,5,5,1,1,2,)
\newline 
or, if you run some command, like read.table(), you can save its result as
 a data frame named 
\begin_inset Quotes eld
\end_inset 

b
\begin_inset Quotes erd
\end_inset 


\newline 
> b 
\begin_inset Formula $<-$
\end_inset 

read.table(
\begin_inset Quotes eld
\end_inset 

myData.txt
\begin_inset Quotes erd
\end_inset 

,header=T)
\newline 
Note the equal sign is used inside the parentheses.
 The equal sign is OK there, because it is not making a permanent assignment
 to some variable.
\newline 

\newline 
If you study other math-oriented computer packages, such as 
\emph on 
Mathematica
\emph default 
, you will find the same kind of distinction between equals and assignment.
 Its not an R thing in particular.
\layout Enumerate

Terminology:
\newline 
I only bring these things up because you are likely to get confused if you
 go into some manual and find that they are using a lot of terms that are
 unfamiliar to a social scientist.
\begin_deeper 
\layout Enumerate

object: a 
\begin_inset Quotes eld
\end_inset 

thing
\begin_inset Quotes erd
\end_inset 

 that 
\begin_deeper 
\layout Enumerate

has variables (one or more): that means it has a 
\begin_inset Quotes eld
\end_inset 

name
\begin_inset Quotes erd
\end_inset 

 and a 
\begin_inset Quotes eld
\end_inset 

value
\begin_inset Quotes erd
\end_inset 

 for each variable.
 The value can be a missing value.
\layout Enumerate

can do 
\begin_inset Quotes eld
\end_inset 

stuff
\begin_inset Quotes erd
\end_inset 

 that you ask it to.
\end_deeper 
\layout Enumerate

method: the instructions that are sent to objects.
\newline 

\newline 
I frequently betray my SAS background by calling these things 
\begin_inset Quotes eld
\end_inset 

procedures
\begin_inset Quotes erd
\end_inset 

.
\newline 

\newline 
In R, the style is different than I was used to from computer languages
 like Java and Objective-C.
 (It seems to me that, in R, one thinks of a message sent to an object in
 a way that is almost exactly the opposite of Java.) If you ask an object
 to carry out a method, in R you say: 
\newline 

\newline 
> 
\begin_inset Formula $method(object)$
\end_inset 


\newline 

\newline 
whereas in java you would say:
\newline 

\newline 
object.method()
\newline 

\layout Enumerate

In R, you can do things however you like, but there is a certain recommended
 style.
 Most importantly, always remember that the results of methods are always
 objects.
\newline 

\newline 
You can run a regression model and watch the output on the screen with this
 sort of command:
\newline 

\newline 
lm(y~x)
\newline 

\newline 
but most people don't do that.
 Rather, they save the 
\begin_inset Quotes eld
\end_inset 

regression object
\begin_inset Quotes erd
\end_inset 

 and then use it:
\newline 

\newline 
myRegObj <- lm(y~x)
\newline 
summary(myRegObj)
\newline 

\newline 
summary is the method, which the object myRegObj carries out.
 The R style looks like a function call, but it is not, really.
\newline 

\newline 
You see my Objective-C background pop out here because I like to name things
 verbosely, starting with a small letter, then using capitals to start new
 elements of the name.
\end_deeper 
\layout Enumerate

In R, you can type in lots of stuff.
 But you should not.
 
\newline 

\newline 
Write your commands into a text editor and then run them in R.
 This way, you have a perfect record of what you have done and you can always
 reproduce it perfectly.
\newline 

\newline 
The editor Emacs (and Xemacs) can be customized to work together with R,
 but I don't think there is a big advantage for most users from doing that.
\layout Enumerate

R has what our students need.
\begin_deeper 
\layout Enumerate

The standard R distribution includes all statistical methods (and many more)
 that are used in an intermediate course like POLS 707.
 For linear models, use lm().
 For scatterplots, use plot().
 For histograms, start with hist() (or truehist()).
 For logit models, use glm().
 For factor analysis, use aov().
\layout Enumerate

R includes a general purpose programming language, featuring vectors, matrices,
 and many other excellent things.
 There is no limit to the complexity of your projects (except the limits
 imposed by your patience, intelligence, and access to hardware).
\end_deeper 
\layout Enumerate

Factors!
\newline 

\newline 
R methods are (usually) sensitive to the kinds of variables you ask them
 to use.
 They will automatically treat categorical variables differently from continuous
 variables.
 
\newline 

\newline 
Understand this term:
\newline 

\series bold 
Factor.

\series default 
 S/R terminology for a variable that is not continuous.
 Factors are 
\begin_inset Quotes eld
\end_inset 

grouping variables
\begin_inset Quotes erd
\end_inset 

 and they can be ordered (some, lots, all) or unordered (male, female).
\newline 

\newline 
In the 
\begin_inset Quotes eld
\end_inset 

old days
\begin_inset Quotes erd
\end_inset 

, the statistical software would not try to separate variables that are
 meaningfully scaled (age, income, etc) from variables that were not.
 The user had the duty of creating a coding scheme for categorical variables.
 Remember 
\begin_inset Quotes eld
\end_inset 

recoding
\begin_inset Quotes erd
\end_inset 

 simply to force categorical variables into a numerical framework, so they
 could be put into models?
\newline 

\newline 
Example 1: dem = 1 if the respondent is a Democrat, 0 otherwise,
\newline 
Example 2: vote = 1 if the respondent voted, 0 otherwise.
\newline 

\newline 
In R, if you input some variable with values like 
\begin_inset Quotes eld
\end_inset 

D
\begin_inset Quotes erd
\end_inset 

 and 
\begin_inset Quotes eld
\end_inset 

R
\begin_inset Quotes erd
\end_inset 

 and then you declare it as a factor, then R will 
\begin_inset Quotes eld
\end_inset 

automagically
\begin_inset Quotes erd
\end_inset 

 recode it for you (turn it into 0 and 1).
 There are a few different ways it can automagically recode categorical
 variables.
 Look for 
\begin_inset Quotes eld
\end_inset 

contrasts
\begin_inset Quotes erd
\end_inset 

 in the documentation.
 If a model does not make sense, then R will tell you so.
\newline 

\newline 
Check this out:
\newline 
> sex <- c("M","F","M","F","M","F","M","F"
\newline 
> age <- c(2,3,1,4,2,4,2,1) 
\newline 
> #lm(age~sex) returns an error, but this does not
\newline 
> sex <- as.factor(sex) 
\newline 
> lm(age~sex) 
\newline 

\newline 
The lm method automagically creates a dummy variable 
\begin_inset Quotes eld
\end_inset 

sexM
\begin_inset Quotes erd
\end_inset 

 that is used in the regression model 
\begin_inset Formula $y=\beta_{0}+\beta_{1}sexM$
\end_inset 

.
 
\newline 

\newline 
Now change sex:
\newline 
> sex <- c("M","F","M","F","M","F","I","I") 
\newline 
> sex <- as.factor(sex)
\newline 
> lm(age~sex)
\newline 

\newline 
That automagically creates 2 dummy variables, 
\begin_inset Quotes eld
\end_inset 

sexM
\begin_inset Quotes erd
\end_inset 

 and 
\begin_inset Quotes eld
\end_inset 

sexI
\begin_inset Quotes erd
\end_inset 

 and it estimates 
\begin_inset Formula $y=\beta_{0}+\beta_{1}sexM+\beta_{2}sexI$
\end_inset 


\newline 

\newline 
If you already have the numerical data, you tell R to use it as a factor
 with the 
\begin_inset Quotes eld
\end_inset 

factor
\begin_inset Quotes erd
\end_inset 

 method and then you can set the labels with the 
\begin_inset Quotes eld
\end_inset 

levels
\begin_inset Quotes erd
\end_inset 

 method.
 
\newline 
> sex <- c(0,1,0,1,0,1,0,1)
\newline 
> factorSex <- factor(sex,levels=0:1)
\newline 
> levels(factorSex) <- c("M","F","I") 
\newline 
> factorSex 
\newline 
[1] M F M F M F M F 
\newline 
Levels: M F I 
\newline 

\newline 
I do not know if S was the first language to explicitly build in the cautious
 treatment of categorical variables, but I'm pretty sure it was before SAS
 in that regard.
 Now SAS has the 
\begin_inset Quotes eld
\end_inset 

CLASS
\begin_inset Quotes erd
\end_inset 

 option in some procedures.
 It works like factors in R.
\layout Enumerate

Graphics: The agony and the extacy.
\begin_deeper 
\layout Enumerate

Device.
 This terminology baffled me for a long time.
 I tend to think of device as a machine.
 But that's not what R means by device.
 Device means an output format.
 The screen output device x11().
 If you want to write the putput in a postscript file, use postscript().
 There are devices for jpg, png.
 In MS Windows, there is a device to save windows meta file format.
\layout Enumerate

In Rtips I give some details about saving graphs.
 I think it is a pain, frankly, but here is the story.
 If you make a graph on the screen, you can't always save it into a file
 so it looks just the same.
 To be save, it is necessary to fiddle the graph on the screen, and then
 turn on an output device that writes a file, and then re-run the graph
 command.
\end_deeper 
\layout Enumerate

There is some stuff in R-base, but there is much much more in R addon packages.
 If you want to use those packages, you have to explicitly load them with
 the library() command.
\newline 

\newline 
R is an 
\begin_inset Quotes eld
\end_inset 

extensible
\begin_inset Quotes erd
\end_inset 

 product, meaning that users can freely write additional capability, package
 it up, and distribute it.
 When I first heard of R, there was complete 
\emph on 
package anarchy
\emph default 
.
 Now there is a recommended set of packages that the R makers put together
 with R when it is distributed, and on the R Internet nexus called 
\begin_inset Quotes eld
\end_inset 

CRAN
\begin_inset Quotes erd
\end_inset 

 one can find many more packages that work, to varying degrees.
\newline 

\newline 
Example.
 
\begin_inset Quotes eld
\end_inset 

truehist
\begin_inset Quotes erd
\end_inset 

 shows up in the output from help.search(
\begin_inset Quotes eld
\end_inset 

hist
\begin_inset Quotes erd
\end_inset 

):
\newline 
truehist(MASS) Plot a Histogram 
\newline 
This means that the package(MASS) (guess what that's short for) has another
 histogram method.
 In order to use that, you must load the package with this command
\newline 
> library(MASS)
\newline 
And then you can read the help page:
\newline 
> ?truehist
\newline 
In that writeup, you see that you get access to many more settings than
 you get with the ordinary hist method.
\newline 

\newline 
Here are packages that I will install and will be available in our stat
 lab
\begin_deeper 
\layout List
\labelwidthstring 00.00.0000

car: a companion for applied regression by John Fox, who has published a
 several textbooks on regression modeling, the most recent of which is from
 Sage, 
\emph on 
R and S+ Companion to Applied Regression
\emph default 
.
\layout List
\labelwidthstring 00.00.0000

Rcmdr: a gui interface for R, also by John Fox.
 Type
\newline 
> library(Rcmdr)
\newline 
to start it up.
 It is being actively updated.
 I notice it gets better all the time.
 Also from John Fox.
\layout List
\labelwidthstring 00.00.0000

mgcv: a package for generalized additive regression and Smoothing models
\layout List
\labelwidthstring 00.00.0000

Hmisc,Design: packages from Frank Harrell, a biomedical stats professor/research
er who previously had written procedures for SAS and S+.
\layout Standard

I will install other packages if you ask nicely and I agree they are needed,
 otherwise you can install packages in your private user space.
 R has documentation for how to do that.
\end_deeper 
\layout Enumerate

Model notation in R.
 The literature on the 
\begin_inset Quotes eld
\end_inset 

generalized regression model
\begin_inset Quotes erd
\end_inset 

 introduced a streamlined notation for regression models that is followed
 in R and some other languages.
 It is called 
\begin_inset Quotes eld
\end_inset 

Wilkinson and Rogers
\begin_inset Quotes erd
\end_inset 

 notation and R's notation is very close to W&R.
 It is described on p.
 75-79 in 
\emph on 
An Introduction to R.
 
\emph default 
If you want to estimate a model:
\begin_inset Formula \[
y_{i}=\alpha+\beta_{1}x1_{i}+\beta_{2}x2_{i}\]

\end_inset 

 Then the formula notation for that in R is
\begin_inset Formula \[
y\sim x1+x2\]

\end_inset 

 Typically, the software will assume you want an estimate for the constant
 (
\begin_inset Formula $\alpha)$
\end_inset 

 and that you want to estimate separate coefficients (
\begin_inset Formula $\beta_{1}$
\end_inset 

and 
\begin_inset Formula $\beta_{2})$
\end_inset 

 for varables 
\begin_inset Formula $x1$
\end_inset 

and 
\begin_inset Formula $x2$
\end_inset 

.
 
\newline 

\newline 
Suppose you tell R you want to estimate the formula:
\begin_inset Formula \[
y\sim x1*x2\]

\end_inset 

 That will estimate a model that includes independent variables 
\begin_inset Formula $x1$
\end_inset 

 and 
\begin_inset Formula $x2$
\end_inset 

 as well as the 
\begin_inset Quotes eld
\end_inset 

interaction
\begin_inset Quotes erd
\end_inset 

 
\begin_inset Formula $x1*x2$
\end_inset 

:
\begin_inset Formula \[
y_{i}=\alpha+\beta_{1}x1_{i}+\beta_{2}x2_{i}+\beta_{3}x1_{i}*x2_{i}\]

\end_inset 

 If you happened to enter a formula like 
\begin_inset Formula $x1*x2*x3*x4*x5,$
\end_inset 

R would estimate a LOT of coefficients, because you'd get an estimate for
 each variable, each pair of variables multiplied together, each set of
 3 variables multiplied, and so forth.
\newline 

\newline 
There is another especially convenient element in R.
 You can use any of the mathematical functions that R has 
\emph on 
inside
\emph default 
 your formula.
 For example, a formula like:
\newline 

\newline 
> logReg <- lm (y ~ x1+log(x2))
\newline 

\newline 
will provide estimates for a model:
\newline 

\begin_inset Formula \[
y_{i}=\beta_{0}+\beta_{1}*x1+\beta_{2}*log(x2)\]

\end_inset 


\newline 
I posted an example program 
\begin_inset Quotes eld
\end_inset 

easyLogReg.R
\begin_inset Quotes erd
\end_inset 

 to demonstrate that.
 You can find it in my R ExampleCode directory
\newline 
http://www.ku.edu/~pauljohn/R/ExampleCode
\layout Enumerate

With R, you can do many things that other stat frameworks would not allow
 or facilitate.
\begin_deeper 
\layout Enumerate

In Stata and SPSS, you can only open one data set at a time.
 That's a crippling limitation, in my opinion.
\layout Enumerate

R is similar to SAS in the sense that one can run a model, save its results
 into various new datasets, and then do additional work with those new datasets.
 However, R is orders of magnitude more convenient.
 Compare the R code to illustrate the Central Limit Theorem with the mean
 of a Gamma variable
\newline 
http://www.ku.edu/~pauljohn/R/ExampleCode/template_gamma4.R
\newline 
against the SAS code for the same purpose:
\newline 
http://lark.cc.ku.edu/~pauljohn/SASClass/ExampleCode/clt-gamma2.sas
\layout Enumerate

R has great facilities to handle repetitive tasks.
 I have run simulations that generate hundreds of output data sets.
 I have written R code that can systematically open each one, make plots,
 and calculate results.
 If I had to do that by hand, I'd be as insane as most Windows users.
\end_deeper 
\layout Enumerate

You can join the r-help email list and you can ask for help in there.
 Heed my warning.
 Many people will be helpful to you, but if your question betrays a total
 lack of effort to read the easily available documentation, then they will
 not be so eager to help.
 
\newline 

\newline 
I find the r-help list is most helpful for relatively specific questions
 about how some specific command R can be used or a specific malfunction
 that you have observed.
 Nobody is interested in a general email like 
\begin_inset Quotes eld
\end_inset 

I can't make linear models work in R
\begin_inset Quotes erd
\end_inset 

, but they are eager to help if you ask a specific question about the syntax
 in the method lm().
 They are also eager to point your attention to packages that provide certain
 functions, but they'll urge you to search on CRAN before asking.
 For example, if you search on CRAN and see 4 packages that can handle 
\begin_inset Quotes eld
\end_inset 

multiple imputation for missing data,
\begin_inset Quotes erd
\end_inset 

 it would be suitable to join r-help and ask if anybody will share their
 experiences with thes packages.
\newline 

\newline 
If you want to ask in the r-help email list, you should follow this progression.
\begin_deeper 
\layout Enumerate

Run help.search(
\begin_inset Quotes eld
\end_inset 

your topic
\begin_inset Quotes erd
\end_inset 

) to see what pops up.
\layout Enumerate

Consult 
\begin_inset Quotes eld
\end_inset 

Introduction to R,
\begin_inset Quotes erd
\end_inset 

 
\layout Enumerate

Check the Venables & Ripley book.
 Check the Daalgard book.
\layout Enumerate

Look through my Rtips collection of usage tidbits.
 That's what I do, and when I learn something new, I often put it in there.
 Or at least I did a while ago:
\begin_inset LatexCommand \url[Rtips:]{http://www.ku.edu/~pauljohn/R/Rtips.html}

\end_inset 


\layout Enumerate

Go to the R email list online archive and search for some key words.
 Chances are, if you have a question, some other person already had it.
 You can find about the R mailing lists and their archives on the main R
 web site: 
\begin_inset LatexCommand \url{http://www.r-project.org/}

\end_inset 

.
 If you go to this page, which also keeps email list archives, 
\newline 

\begin_inset LatexCommand \url{http://maths.newcastle.edu.au/~rking/R/}

\end_inset 


\newline 
at the bottom of the page you will find a SEARCH tool that is rather convenient.
\the_end
e_end