The R Book 2nd Edition Pdf Download
Discover the world's research
- 20+ million members
- 135+ million publications
- 700k+ research projects
Join for free
JSS
Journal of Statistical Software
October 2015, Volume 67, Book Review 5. doi: 10.18637/jss.v067.b05
Reviewer: James E. Helmreich
Marist College
Statistics: An Introduction Using R (2nd Edition)
Michael J. Crawley
John Wiley & Sons, Chichester, 2015.
ISBN 978-1-118-94109-6. 339 pp. USD 45.00 (P).
http://www.imperial.ac.uk/bio/research/crawley/statistics/
Introduction
Michael J. Crawley's text Statistics: An Introduction with R is indeed an introduction to
statistical analysis as well as excellent introduction to working in and with R. However, it is
not for the faint of heart. The first half covers standard introductory material (descriptive
statistics, simple regression and inferential statistics) in a somewhat idiosyncratic way. The
second half of the text is a good introduction to modeling (multiple regression, ANOVA and
ANCOVA, various general linear models, survival analysis). The treatments are thorough
and yet short – only 300 odd pages. As you can imagine he is extraordinarily succinct. There
are many useful and thoughtful insights; it is well written and clear. It contains no exercises;
there is an excellent appendix that serves well as a basic R tutorial. If you teach this material
without a text or think you might like to try, you might find that it (paradoxically) will work
as the structural backbone for a course that you fill out as you see fit. The text will serve
well for an introductory course or for a second course in modeling.
Discussion
The overall structure and approach is very nice. His use of R is quite gentle, intertwined
with the narrative in a way that is natural, informative, and quickly builds basic facility. In
early chapters Crawley introduces a statistic, then calculates it directly 'by hand' in R. He
writes a function to automate the process, and only having done so points out the built-in R
version. He simultaneously discusses the meaning and use of the statistic, code to calculate it,
and higher level R structures. For instance, within the first few pages using R he introduces
functions, modular arithmetic, nested looping and ways to plot the results. This sounds like
a lot and it is, but it works well here. The magical black-box feel of many routines in varied
platforms should be dispelled by the approach. Once in a while he does not follow through –
2 Statistics: An Introduction Using R (2nd Edition)
e.g., deriving the least squares coefficients yet going directly to lm – but for the most part he
uses this method to good effect.
The material is kept quite applied. No real discussion of the mathematics or theory is provided.
Nevertheless he is careful with algebraic arguments concerning partitioning sums of squares or
calculation of degrees of freedom; he derives the (two-variate) normal equations and solutions.
These types of calculations are the only instances of boxed, independent asides outside of the
general narrative.
The book starts out with a chapter of fundamentals, more of an annotated glossary for
later reference than an introduction. Here we find general guidelines for types of analyses;
discussion of hypotheses and error types; typical assumptions; experimental design; maximum
likelihood estimation; power; replication; randomization; weak and strong inference; multiple
comparisons; as well as a laundry list of some of the basic model types available in R. In
twenty-two pages. Whew!
The presentation is rather terse – the subsection on controls reads in its entirety:
No controls, no conclusions.
Unfortunately it is often opaque at first (student) reading:
Pseudoreplication occurs when you analyze the data as if you had more degrees
of freedom than you really have.
This is the first mention of degrees of freedom I can find in the text; these sorts of out of
place sections lard the text, and can cause some confusion. While frustrating to students
new to the material, it is in service of brevity, elegance of presentation, and perhaps piquing
the interest of the reader for what is to come. Though at times annoying, I find it works
reasonably well overall. Clearly, it is intended that some of the material will become clearer
as the reader progresses through the book.
The second chapter begins the material on R, introducing dataframes and their manipulation,
as well as "initial data inspection" – summary statistics and basic graphical analysis (in that
order: I'd reverse it, but then that's me). Datasets from the book's website are used, with R
commands highlighted in red, the output in blue.
The third chapter begins where most texts start: measures of central tendency. Here, as well
as in later chapters, the treatment is not completely standard. A discussion of quartiles and
the five number summary is omitted. After discussing the mean and median, the geometric
and harmonic means are presented (with good motivating examples) as well.
Another instance of non-linearity: the next chapter on variance ends with a residual scatter-
plot showing heteroscedasticity long before chapters on regression or bivariate data. I found
myself frequently wishing for a more thorough index. On the other hand, Crawley introduces
confidence intervals very naturally and intuitively, both t and bootstrapped versions. He
demonstrates sample size effects on estimation using nested loops for sampling and graphics.
It is an odd, or at least unorthodox, presentation that works nicely.
In the next chapters he introduces analysis of single samples and two samples, including nor-
mality tests and both parametric and nonparametric inference techniques. The single-sample
chapter finishes up with a discussion of statistics to measure skew and kurtosis. Two-sample
nonparametric alternatives (rank sum and binomial test) feature prominently. He includes
Journal of Statistical Software – Book Reviews 3
graphical techniques as well: to with notched box plots. The χ
2
and Fisher's exact tests
for independence of two samples of counts is covered. The two-sample discussion concludes
with correlation – including Pearson's product-moments correlation method of testing for
differences from zero.
From here, Crawley begins simple regression by reminding the reader of the definition of slope
and other rather rudimentary topics. Curiously, he spends a rather verbose three pages at a
mathematical level dramatically lower than the rest of the text's narrative. That aside, the
presentation is very good, and quickly goes well beyond normal material to include partition-
ing sums of squares; calculating standard errors for the slope and intercept, careful model
checking and transformations to correct problems, polynomial regression and even non-linear
regression as well as generalized additive models.
Again, whew! This recalls to mind a colleague's quip that assigning a difficult text is a
desirable thing in that it allows you to be the good guy when you elaborate and clarify in
the classroom. Too many modern texts are written in such a way that the roles are reversed.
But I overstate: the worked examples are clear, interesting and easily replicated with the
commands shown. Color is used well but sparingly to differentiate text, code, and output, and
for graphics. There are no colored boxes, marginal notes, distracting highlights or gratuitous
photos in the text.
The narrative continues with chapters on ANOVA, ANCOVA, multiple regression, contrasts,
and a series of chapters on general linear models. In all of these the focus is on the nature of
the data. Poisson regression is discussed in the chapter on count data, logistic in the chapters
on proportion data and binary response data. A brief chapter on survival analysis rounds out
the text.
These regression chapters are my favorite part of the book. With minor variations each
chapter in the latter half of the book takes a data set, sets out the situation, does a basic
analysis, finds a maximal model and then either by hand or using step prunes to a minimal
adequate model. The implications of the model and need for transformations are assessed,
and if necessary the process is repeated. He uses graphics well to assess models and composes
nice multilayer plots of the data and fitted curves.
The entire set of chapters begins with an overview of types of data, error structures, link
functions, deviance and AIC. This type of overview is one of the great strengths of the
exposition. Crawley provides excellent characterizations at various points outlining the big
picture or the general steps an analysis should take.
I have a few quibbles. The exposition is, again, at times non-linear. We frequently meet terms
that are only defined later in the text. There is very little theory, which gives many of the
topics a rabbit-out-of-the-hat feel to them – though when possible nice intuitive comments
(e.g., for maximum likelihood techniques) are presented in lieu of the mathematics.
There are some inconsistencies. He makes a strong case for graphical analysis of regression
results, providing an example of a model with a strong numeric summary that nevertheless
graphical analysis shows to be flawed. The point is a good one if not consistently followed.
Before this example there were several occasions where he proceeds, willy-nilly, with a numer-
ical analysis without even a precaution to the student that they consider graphing the data
first. Indeed, in Chapter 4 variance is calculated repeatedly while ignoring any plots.
4 Statistics: An Introduction Using R (2nd Edition)
Conclusion
Crawley has a real knack for outlining the general steps and processes of an analysis that is
clear, concise, informative and useful. He frequently makes pithy, insightful points without
overelaboration. He steps back regularly to give a broad overview of the topic. The use of R is
integrated seamlessly into the exposition, and will provide the novice with a straightforward
introduction to R code. The approach is not quite what I expected from an introductory text,
but I am left with the strong sense that it should be.
I could easily see this used as the backbone of both a first-year course for STEM students
with the mathematical maturity to take Calculus, as well as a course in applied statistics for
students who have taken at least a semester of mathematical probability and statistics. It
does omit exercises, and given the author's brevity will need elaboration in the classroom.
However the text provides strong overall structure and would be a good student resource.
Having taught the material from the second half of the text recently without a text, I find
myself wishing I had known of and made this text available to my class.
Reviewer:
James E. Helmreich
Marist College
Department of Mathematics
3399 North Road
Poughkeepsie, NY 12514, United States of America
E-mail: James.Helmreich@Marist.edu
URL: http://foxweb.marist.edu/users/james.helmreich/
Journal of Statistical Software http://www.jstatsoft.org/
published by the Foundation for Open Access Statistics http://www.foastat.org/
October 2015, Volume 67, Book Review 5 Published: 2015-01-06
doi:10.18637/jss.v067.b05
... The different cognitive, psychomotor and, emotional development is certainly a consideration in choosing elicitation techniques. Interview and prototype techniques [21], are recommended as appropriate techniques for child respondents. The selection of appropriate techniques needs to be followed by appropriate tools so that communication can go well. ...
... Meanwhile, to achieve the second goal, 32 participants were involved. The number of participants is adjusted to the limit of quantitative research [21], as many as 30 children. ...
... Hence there were no potentially problematic outliers detected. In addition, constant variance (homoscedasticity) and normally distributed errors are two essential assumptions of any linear model, including linear fixed effects panel models (Helmreich, 2015). The regression assumptions were evaluated, and the residuals were normally distributed. ...
This study examines the impact of increasing climate variability on food production in South Africa, focusing on maize and wheat yields. A two-way fixed effects panel regression model was used to assess the climate variability impacts, analysing secondary data for the period 2000 to 2019 for nine provinces in South Africa. The study found that increasing climate variability has a negative impact on maize and wheat production in South Africa. Specifically, the results indicated a negative correlation between mean annual temperature with both maize and wheat yields. A decrease in precipitation affected maize yields negatively, while the impact on wheat yields was positive, although insignificant. This analysis, therefore, depicted that crop yields generally increase with more annual precipitation and decrease with higher temperatures. The study recommends that funding initiatives to educate farmers on increasing climate variability and its effects on farming activities in South Africa should be prioritised.
... The p-value for all the variables used in the panel regression models was found to be greater than 0.05, hence there were no potentially problematic outliers detected. In addition, constant variance (homoskedasticity) and normally distributed errors are two essential assumptions of any linear model, including linear fixed-effects panel models (Helmreich, 2015). The regression assumptions were evaluated, and the residuals are normally distributed. ...
This study examines the impacts of increasing climate variability on food production in South Africa, focusing on maize and wheat yields. A two-way fixed-effects panel regression model is used to assess the climate variability impacts, analysing secondary data for the period 2000 to 2019 for nine provinces in South Africa. The study finds that the increasing climate variability has a negative impact on the production of maize and wheat in South Africa. Specifically, the results indicated a negative correlation between mean annual temperature with both maize and wheat yields. A decrease in precipitation affected maize yields negatively, while the impact on wheat yields was positive, although insignificant. This analysis therefore depicted that crop yields generally increase with more annual precipitation and decrease with higher temperatures. The study recommends that funding initiatives towards educating farmers on increasing climate variability and its effects on farming activities in South Africa should be prioritised.
Worldwide urban expansion and deforestation have caused a rapid decline of non-human primates in recent decades. Yet, little is known to what extent these animals can tolerate anthropogenic noise arising from roadway traffic and human presence in their habitat. We studied six family groups of titis residing at increasing distances from a busy highway, in a park promoting ecotourism near Santa Cruz de la Sierra, Bolivia. We mapped group movements, sampled the titis' behavior, collected fecal samples from each study group and conducted experiments in which we used a mannequin simulating a human intrusion in their home range. We hypothesized that groups of titi monkeys exposed to higher levels of anthropogenic noise and human presence would react weakly to the mannequin and show higher concentrations of fecal cortisol compared with groups in least perturbed areas. Sound pressure measurements and systematic monitoring of soundscape inside the titis' home ranges confirmed the presence of a noise gradient, best characterized by the root-mean-square (RMS) and median amplitude (M) acoustic indices; importantly, both anthropogenic noise and human presence co-varied. Study groups resided in small, overlapping home ranges and they spent most of their time resting and preferentially used the lower forest stratum for traveling and the higher levels for foraging. Focal sampling analysis revealed that the time spent moving by adult pairs was inversely correlated with noise, the behavioral change occurring within a gradient of minimum sound pressures ranging from 44 dB(A) to 52 dB(A). Validated enzyme-immunoassays of fecal samples however detected surprisingly low cortisol concentrations, unrelated to the changes observed in the RMS and M indices. Finally, titis' response to the mannequin varied according to our expectation, with alarm calling being greater in distant groups relative to highway. Our study thus indicates reduced alarm calling through habituation to human presence and suggests a titis' resilience to anthropogenic noise with little evidence of physiological stress. How to cite this article Hernani Lineros LM, Chimènes A, Maille A, Dingess K, Rumiz DI, Adret P. 2020. Response of Bolivian gray titi monkeys (Plecturocebus donacophilus) to an anthropogenic noise gradient: behavioral and hormonal correlates. PeerJ 8:e10417 http://doi.
ResearchGate has not been able to resolve any references for this publication.
Posted by: fitathomeand.blogspot.com
Source: https://www.researchgate.net/publication/282899937_Statistics_An_Introduction_Using_R_2nd_Edition
