Biostat/Stat 572 and Stat 518
- Acknowledgement: --- Ken Rice!
- Week 1
- 1 April 2014: Course overview / searching the literature (Patrick out)
- J. Wakefield: Course Rationale and Expectations
- P. Heagerty: Research / Reading the Literature
- Murphy (2007) --- "How to
Read the Statistics Methodology Literature - A Guide to Students",
The American Statistician
- Hamada and Sitter (2004)
--- "Statistical Research: Some Advice for Beginners",
The American Statistician
- Schwartz (2008) --- on why not knowing everything is an important feeling when doing research.
- Marie Davidian's NCSU course on how to do research
- Korner: How to Write a Part III essay. These are notes for University of Cambridge math postgrads, who (like you) have to write a report based on published literature.
- Links to online citation databases, using your UW access;
- Web of Knowledge, for abstracts, citations, references and links to papers. Covers most scientific areas
- Current Index to Statistics, for bibliographic information in BibTeX format. Only statistics articles, and some books
- Google Scholar, for citations, multiple links to papers, and 'related articles'. Get BibTeX citations by setting your 'Google preferences'
- PubMed, for abstracts, citations, references and links to papers. Covers NIH-related areas
- arXiv, open access to e-prints in Physics, Mathematics, Computer Science, Quantitative Biology, Quantitative Finance and Statistics
- MathSciNet, American Mathematical Society (AMS) search tool
- Q: what are the most highly cited
statistics papers for each of the last few decades?
- Assignment: --- preparation for Thursday:
- Choose (2) papers from the "Presentation Papers" listed below
- Send me e-mail by Weds 5pm telling me the papers that you have
- Using your paper identify (3) guidelines for a good talk --
use the themes of ORGANIZATION, CONTENT, and PRESENTATION to
label your guidelines if possible.
- Review the "Ken's 2011 slides" below to gain another
perspective -- are your identified guidelines apparent in
- Goal: To formulate a set of GUIDELINES and SUGGESTIONS.
- Be prepared to discuss results on Thursday.
- 3 April 2014: Preparing and delivering a presentation (Jon out)
- M. Meila: Beamer / tools
Ken's 2011 slides (PDF)
- From Navneet H: Numerator Only! -- how to only display slide counter and not total
Slides and other material on How To Communicate Graphically
For a primer on drawing graphs in R, see this course
An example Beamer file, with figures (ZIP file) - intended to be self-explanatory. Made using PCTeX, it may require minor tweaks in order to work on other systems
A LaTeX example featuring DAGs - taken from Ken's 570 notes
- A link to Tikz is HERE
- Example from WenWei Loh:
- Simple example of using pstricks to create figure:
- From Adam G:
igraph (note that this does a lot more than having some nice plotting algorithms)
- From Adam G:
Rgraphviz (note that this requires installing the graphviz software on your system)
- Presentation Papers
Some examples of good speakers;
An outstandingly good 20 minute talk; Hans Rosling (Swedish Statistician) explains data visualization using mortality vs life expectancy, in 200 countries over 200 years, as a 4-minute example. Your talks will be less of a sales pitch, and aimed at a more expert audience - but think about the way he is keeping the audience engaged
A few minutes of Richard Feynman, explaining some connections between Physics and Math. His example is extremely clear, but also note how he paces the material for the audience - not himself
Two similar talks (both are good); Peter Donnelly explains the prosecutor's fallacy to a non-expert audience. David Spieglhalter discusses the same idea for an even-less-expert audience, but using more graphics - and in a lot less time
- Assignment: --- for your selected paper:
- Search the bibliography of your paper and list the (3) most
highly cited papers that your paper references. (and note the
number of citations for each).
- Summarize the papers that have cited your project paper. Are
there themes (methodologically, scientifically) that are
- Also look at citations in the "second generation" -- e.g.
citations among papers that cite your project paper. Just
choose a couple and explore the downstream impact of your paper.
- Be prepared to discuss results on Tuesday.
- Week 2
- 8 April 2014: Writing Statistical Papers
- M. Meila: Calling C from R, and Cplex optimization.
- P. Heagerty: Statistical papers
- J. Wakefield: LaTeX paper format
- Ken's 2011
Slides (PDF) and a zip file with a BibTeX example
Google Scholar can output BibTeX-format bibliography entires; just set your 'Scholar Preferences', hit 'Import into BibTeX', then cut and paste
Also, Zotero is a (free) Firefox add-in that allows straightforward "scraping" of citation information from websites. Having done this, it formats the data for use in other software, e.g. BibTeX
Gopen and Swan (1990) The Science of Scientific Writing is a classic paper on writing clearly
UChicago offers a Sentence of the Week, with great examples of how to make writing clearer. Their writing resources are also good
Strunk and White The Elements of Style is a classic style guide. 'Classic' also means 'old', so don't treat it as gospel - and it is far from perfect. Also, technical writing is often slightly more formal than their 'style'.
The Chicago Manual of Style is also a standard reference, see also The Complete Plain Words
Get easily-digested style tips from this NY Times blog
Paul Brians has an excellent list of non-errors about which grumpy trolls may be picky. His list of real errors is also a useful reference
Williams and Colomb's point about 'Noun + Noun + Noun' was made (grumpily, but in Science) by Milton Hildebrand (1983). He was subsequently criticized by Baer (1983), who recommended using hyphens where possible
Some fun stuff; TheOatmeal on;- thanks to Luis Crouch for these
A fairly definitive note on why to use only one space after a period... which is what LaTeX will do for you automatically
The Mayfield Handbook of Scientific and Technical Writing is a specialized style guide
A checklist of Word Usage in Scientific Writing
Twenty one Suggestions for Writing Good Scientific Papers
Ehrenberg 1982 Writing Technical Reports or Papers (American Statistician) gives good advice on structure, and use of simple language
Andrew Gelman reviews two research articles
Halmos' classic advice on How To Write Math
Lee Notes on Writing Mathematics (from the UW Math Dept)
Knuth, Larrabee and Roberts (1987) Mathematical Writing. This lengthy report also came out as a book
Higham (1998) Handbook of Writing for the Mathematical Sciences, Second Edition, SIAM (partly-viewable through Google Books)
- Krantz (2000). A Mathematician's Survival Guide. American Mathematical Society.
- Krantz (2000). Mathematical Publishing: A Guidebook. American Mathematical Society.
- Krantz (1991). A Primer of Mathematical Writing. American Mathematical Society.
- 10 April 2014: Intro Student Presentation  (Overview : Motivation
/ Background / Methods intro)
- These talks should be 20 minutes including 5 minutes of questions.
- The goals of the introductory presentation are to introduce the
paper and to motivate the work. A key part of any talk/paper is
showing the audience the need for new methods/theory. This
typically involves both scientific motivation using an example
or data set, and statistical motivation indicating that no
adequate solution exists (e.g. the need for something new!).
Basic elements of the first talk could include:
- Introduce the paper
- Provide motivation for the work (scientific, statistical)
- Review key background literature
- Overview of the methods
- Aaron Baraff Talk
- Amrit Dhar Talk
- Rebecca Ferrell Talk
- Week 3
- 15 April 2014: Intro Student Presentation 
- 17 April 2014: Intro Student Presentation  (Jon gone)
- Week 4
- 22 April 2014 Intro Student Presentation 
- 24 April 2014: Finish Talks and More on writing / editing
[ Draft Due: 6 May Intro and Literature Review]
- Your literature review should provide a summary of the major background literature, and should establish motivation for the work in your paper. This may look similar to the lit review in your paper, but your version will probably need to be written for a less-expert audience, e.g. for those who've had the material up to and including the 570-571 (516-517), but are not familiar with other material.
The goal of this exercise is to give you practice in this important part of scholarly research, and to develop your background knowledge of the statistical area of focus.
A key part of this exercise is organizing the material; you may find it helpful to summarize major themes in turn (e.g. Empirical Bayes, then measurement error), and then identify novel relationships between these, and/or gaps in the methodological literature (e.g. use of Empirical Bayes for measurement error problems)
Your review should be between 2-4 pages. Use BiBTeX for reference management and citation within your report.
Bibtex Style Files for Statistics Journals.
Some example literature reviews, from 2010:
- Week 5
- 29 April 2014: Update Student Presentation  (Patrick out)
- These talks should be 20 minutes including 5 minutes of questions.
- The goal of the "update talks" is for the student to make progress on their paper. In the update talks, students can present on issues that have arisen, and we will allow time for discussion of potential solutions. Please edit your first talk (as needed) and then add new slides that describe your progress. Focus on one or more of the following:Your final presentation should include these elements and therefore the update should represent the logical next step in your work. Please aim to take your allocated 15 minutes, and don't feel the need to review any earlier slides beyond the single slide that reminds us of the basic problem.
Details of the methods in terms of derivation and/or implementation
Details of data simulation that could evaluate the methods
Illustration of the methods using a worked example
- Where you are encountering difficulties
- 1 May 2014: Update Student Presentation 
- Week 6
- 6 May 2014: Update Student Presentation  Literature reviews due
- 8 May 2014: Update Student Presentation 
- Week 7
- 13 May 2014: Update Student Presentation 
- 15 May 2014:
- Return literature reviews and make comments
- Simulations and Computing in R (resources) and extreme coding
- Week 8
- 20 May 2014: Final Student Presentation 
These talks should be 20 minutes with an additional 5 minutes of questions.
- 22 May 2014: More on Writing / Editing (Patrick out)
- Week 9
- 27 May 2014: Final Student Presentation 
- 29 May 2014: Final Student Presentation 
- Week 10
- 3 June 2014: Final Student Presentation 
- 5 June 2014: Final Student Presentation