(r59) CourseBios352 < Main < Vanderbilt Biostatistics Wiki

You are here: Vanderbilt Biostatistics Wiki>Main Web>CourseBios352 (revision 59) (raw view)~~EditAttach~~

---+ Statistical Collaboration Health Sciences II

| Instructors | MarioDavidson MattShotwell |
| Course Number | 2014.01.SPR.GS.BIOS.352 |
| Meeting Times | Monday 1:00pm - 3:00pm |
| Course Location | 2525 W. End Ave; 11139 |
| Office Hours | by request |
<br />%TOC%

---++ Course Description
   * Goal: To develop the knowledge, skills, attitude, and behavior necessary to interact with research collaborators and statistical clients.
   * Each course session will take one of two alternating types. The first type will concern real world consulting or expert input on specialized statistical topics. These sessions will be led by Dr. Davidson (abbreviated "MD" in the schedule below). The second type of session will pursue real consulting problems, including their technical aspects. These sessions will be led by Dr. Shotwell (abbreviated "MS" below).
   * Expectations: Students are expected to participate in classroom discussions, projects, and presentations. Because participation is a portion of the grade, students are expected to attend all sessions. The instructors should be informed beforehand if a student is unable to attend class.
   * All forms of verbal, physical, and written communication are expected to be professional. Students are strongly encouraged to dress professionally for all of Dr. Davidson's sessions.
---++ Course Material and Resources

Students should bring a laptop to class, and are encouraged to follow along with computer demonstrations. Personal laptops with a statistical software should be brought to every class. Participants are expected to have access to MS Excel and Word (or an equivalent such as Open Office). Microsoft Office products are available at the [[https://softwarestore.vanderbilt.edu/][Vanderbilt Software Store]]. A typesetting language is not mandated but encouraged. Data sets may be found in the "External Links" in OAK, on this wiki page, or a sub-page. Writing resources may be found at the [[http://www.vanderbilt.edu/writing][Vanderbilt Writing Studio]]. The studio is free for Vanderbilt University students.
   * _Text_: Statistical Consulting: A Guide to Effective Communication, Janice Derr, ISBN-10: 0534362281, ISBN-13: 978-0534362287
---++ Honor Code

Vanderbilt's Honor Code governs all work in this course. Students may discuss the assignments outside of class, but each student's writing and presentation of the data is expected to be original and independently finalized unless stated otherwise by the instructor. Direct quotes need to cite the source in quotes; otherwise, it is consider plagiarism. If you have any questions about the application of the honor code to this course, please do not hesitate to ask the instructors.
---++ Assignments and Assessment

Assessments and feedback may be found in [[http://www.vanderbilt.edu/oak/][OAK]]. Assignment should be dropped into the "Assignment" folder in [[http://www.vanderbilt.edu/oak/][OAK]] under the appropriate due date and topic. All dropped assignments should be an attachment. Late assignments will be penalized as such: the first day 20%, the second day 40%, and the third day 100%. Assignments may be turned in earlier than the due date. Videos of the consultations may be found at [[BIOS352Videos2014][Consulting Videos]].
   * (45%) Written Statistical Communication Assignments
   * (45%) Statistical Assignments
   * (10%) Biostatistics Clinic Attendance
Numerical scores will be rounded to the nearest tenth. Letter grades will be assigned as defined by the following ranges:
| Letter Grade | Lower Score | Upper Score |
| A+ | 96.5 | 100.0 |
| A | 93.5 | 96.4 |
| A- | 90.0 | 93.4 |
| B+ | 86.5 | 89.9 |
| B | 83.5 | 86.4 |
| B- | 80.0 | 83.4 |
| C+ | 76.5 | 79.9 |
| C | 73.5 | 76.4 |
| C- | 70.0 | 73.4 |
| D+ | 66.5 | 69.9 |
| D | 63.5 | 66.4 |
| D- | 60.0 | 63.4 |
| F | 0.0 | 59.9 |
---++ Schedule
---+++ %BLUE% _Assignment of Roles for Consultations_ %ENDCOLOR%
---++++ Description of Roles

Biostatistician trainees are required to complete multiple projects. A description of the possible roles are below.

Leader
   * Responsible for:
      * managing the project until completion by the deadline(i.e. If a team member is not pulling their weight, the leader is responsible for making sure their part is accomplished.)
      * understanding the topic
      * critically assessing the consultation

Scribe
   * Responsible for:
      * writing the meeting minutes
      * creating a realistic data set and dictionary
      * contacting the investigator (only when necessary on a LIMITED basis)
SAP Writer
   * Responsible for:
      * writing the analysis plan
Analyst
   * Responsible for:
      * cleaning the data
      * analyzing the data,
      * creating graphs
      * creating tables
Final Report Writer
   * Responsible for
      * creating a final report

<table border="0" cellspacing="0"><colgroup width="64"></colgroup> <colgroup width="231"></colgroup> <colgroup width="299"></colgroup> <colgroup width="214"></colgroup> <colgroup span="4" width="85"></colgroup> <tbody> <tr> <td align="center" height="16">Meeting Date</td> <td align="center">Investigator</td> <td align="center">Leader, Background, and Consulting Reflection</td> <td align="center">Scribe, Data Dictionary, and Data</td> <td align="center">SAP Writer</td> <td align="center">Analyst</td> <td align="center">Final Report Writer</td> <td align="center">Deadline</td> </tr> <tr> <td align="center" height="16">02/03/14</td> <td align="center">Charlene Dewey</td> <td align="center">DS</td> <td align="center">CT</td> <td align="center">MZ</td> <td align="center">MG</td> <td align="center">AL</td> <td align="center">02/17/14</td> </tr> <tr> <td align="center" height="16">02/17/14</td> <td align="center">Amy Fleming</td> <td align="center">CT</td> <td align="center">MZ</td> <td align="center">MG</td> <td align="center">AL</td> <td align="center">DS</td> <td align="center">03/10/14</td> </tr> <tr> <td align="center" height="16">03/10/14</td> <td align="center">Regina Russell</td> <td align="center">MZ</td> <td align="center">MG</td> <td align="center">AL</td> <td align="center">DS</td> <td align="center">CT</td> <td align="center">03/24/14</td> </tr> <tr> <td align="center" height="17">03/24/14</td> <td align="center">Tyler Reimschisel</td> <td align="center">MG</td> <td align="center">AL</td> <td align="center">DS</td> <td align="center">CT</td> <td align="center">MZ</td> <td align="center">(First draft) 04/07/14</td> </tr> <tr> <td align="center" height="17"> 04/07/14</td> <td align="center">Oscar Guillamondegui will visit but Dr. Reimschisel's project will be due.  No project will be due for Guillamondegui</td> <td align="center">AL</td> <td align="center">DS</td> <td align="center">CT</td> <td align="center">MZ</td> <td align="center">MG</td> <td align="center">04/21/14</td> </tr> </tbody> </table>
---+++ 2014/01/06 (MD)
   * Ayumi Shintani: Working with, Supervising, and Developing a Team of Biostatisticians
   * Rafe Donahue: Comparing Academic and Industry Biostatisticians
---+++ 2014/01/13 (MS)
   * Select/propose technical topics for future class discussion:
      * Evaluating model building/selection algorithms
      * Combining mathematical and statistical models
      * Web-based reporting (reproducible research) frameworks
      * Double bootstrap ideas
   * Four model selection strategies:
      * "Stepwise"
      * "Bottom Up"
      * "Bottom Step"
      * "Full Model"
   * Link to the figures: https://data.vanderbilt.edu/collab/shotwems/ModelSelection/
   * Code used to draw figures: [[%ATTACHURL%/selection.R][selection.R]]
   * Assignment:
      * Evaluate the "full model" and one other model selection method in an alternative scenario, where the data, modeling approach, and inferential goals are different. You may use the code provided ( [[%ATTACHURL%/selection.R][selection.R]]) as a starting point. Your evaluation need not focus on the distributions of model coefficient estimators, but should be relevant to the scenario. Prepare a reproducible report of your findings.
---+++ 2014/01/20 (MLK Day; No Class)
---+++ 2014/01/27 (MS)
   * last class: S.E. of X1 coefficient depends on selected model
   * last class: selected model depends on sample <p> </p> <p> </p> <p> </p> <p> </p> <p> </p> <p> </p> <p> </p> <p> </p> <p> </p> <p> </p> <p> </p> <p> </p> <p> </p> <p> </p> <p> </p>

<dot>
digraph G {
    Sample -> Model;
    Model -> SE;
    Sample -> SE;
}
</dot>
   * other scenarios (assignment discussions)
   * does X1 coefficient have a consistent inter-model interpretation? Simpson's paradox: <a href="%ATTACHURLPATH%/simpsons-paradox.png">simpsons-paradox.png</a>
   * are most of the models similar, in terms of prediction (MAE)? <a href="%ATTACHURLPATH%/mean-absolute-error-comparison.png">mean-absolute-error-comparison.png</a>
   * conditional on model (Y~X1+X2), are asymptotic S.E. estimates good? <a href="%ATTACHURLPATH%/asymptotic-standard-error-comparision.png">asymptotic-standard-error-comparision.png</a>
   * conditional on model (Y~X1+X2), are bootstrap S.E. estimates good? <a href="%ATTACHURLPATH%/bootstrap-standard-error-comparison.png">bootstrap-standard-error-comparison.png</a>
   * conditional on full model, are asymptotic S.E. estimates good? <a href="%ATTACHURLPATH%/full-asymptotic-standard-error-comparison.png">full-asymptotic-standard-error-comparision.png</a>
   * conditional on full model, are bootstrap S.E. estimates good? <a href="%ATTACHURLPATH%/full-bootstrap-standard-error-comparison.png">full-bootstrap-standard-error-comparison.png</a>
   * Assignment: Read the classic article "The Impact of Model Selection on Inference in Linear Regression" by Hurvich and Tsai (http://www.jstor.org/stable/2685338). Be prepared to discuss its contents at our next meeting. <p> </p> <p> </p>
---+++ 2014/02/03 (MD)
   * William Dupont: Grant Writing as a Collaborator or Primary Investigator.
   * Consultation with Charlene Dewey
---+++ 2014/02/10 (MS)
   * Discuss "The Impact of Model Selection on Inference in Linear Regression" by Hurvich and Tsai (http://www.jstor.org/stable/2685338).
---+++ 2014/02/17 (MD)
   * Consultation with Amy Fleming
   * Elizabeth Heitman: Ethics
---+++ 2014/02/24 (MS)
   * example data: [[%ATTACHURL%/example-data.csv][example-data.csv]] <p> </p> <p> </p> <p> </p>

<verbatim>
dat <- read.csv(paste0("http://biostat.mc.vanderbilt.edu/",
                       "wiki/pub/Main/CourseBios352/",
                       "example-data.csv"))</verbatim>
---+++ 2014/03/03 (Spring Break; No Class)
---+++ 2014/03/10 (MD)
   * Consultation with Regina Russell
   * Stephen Jones: Health Industry Recruiter
   * Rafe Donahue: Comparing Academic and Industry Statisticians (Makeup day: 3-4)
---+++ 2014/03/17 (MD)
   * Teaching Statistics, Discussion about Projects, and/or Student Meeting Time about Projects (TBD)
   * Tatsuki Koyama: Working with, Supervising, and Developing a Team of Biostatisticians (Make-up Day)
---+++ 2014/03/24 (MD)
   * Consultation with Tyler Reimschisel
   * Questionnaire Design
---+++ 2014/03/31 (MS)
   * Collaborative Grant Writing Example: [[%ATTACHURL%/Example.pdf][Example.pdf]]
   * VUMC Grant Writing Resources: http://www.mc.vanderbilt.edu/root/vumc.php?site=ctsd&doc=32383
   * Dept. Biostatistics Guidelines: http://biostat.mc.vanderbilt.edu/wiki/Main/PreStatAnalysisPlan
   * Collaborative Grant Writing Assignment: [[%ATTACHURL%/Biexp.pdf][Biexp.pdf]] [[%ATTACHURL%/Biexp.Rnw][Biexp.Rnw]]
---+++ 2014/04/07 (MD)
   * Creating Effective and Visually Attractive Scientific Posters: Making People Stop!
   * Consultation with Oscar Guillamondegui
---+++ 2014/04/14 (MS)
   * Discuss technical/statistical issues arising from 2013/04/01 consultations
   * More on research proposals
   * Simultaneous bootstrap confidence bands
---+++ 2014/04/21 (MD)
   * Consultation with Leora Horn
   * Brian Griffith: Developing Your Personal Plan
---+++ 2014/04/28 (Finals Week)
   * No Class<br /> <p> </p> <p> </p> <p> </p> <p> </p> <p> </p> <p> </p> <p> </p> <p> </p> <p> </p> <p> </p> <p> </p> <p> </p> <p> </p> <p> </p> <p> </p> <p> </p> <p> </p> <p> </p> <p> </p> <p> </p> <p> </p> <p> </p> <p> </p> <p> </p> <p> </p> <p> </p>

<!-- Set ALLOWTOPICCHANGE = MarioDavidson, MattShotwell -->

Topic attachments
I	Attachment	Action	Size	Date	Who	Comment
pdf	BIOS352FinalProject.pdf	manage	79.7 K	19 Apr 2013 - 10:54	MattShotwell
Rnw	Biexp.Rnw	manage	6.8 K	31 Mar 2014 - 11:32	MattShotwell	Collaborative Grant Writing Template
pdf	Biexp.pdf	manage	90.2 K	31 Mar 2014 - 11:32	MattShotwell	Collaborative Grant Writing Assignment
pdf	Example.pdf	manage	88.6 K	31 Mar 2014 - 11:33	MattShotwell	Collaborative Grant Writing Example
png	asymptotic-standard-error-comparision.png	manage	8.2 K	27 Jan 2014 - 10:57	MattShotwell
png	bootstrap-standard-error-comparison.png	manage	8.2 K	27 Jan 2014 - 10:58	MattShotwell
csv	example-data.csv	manage	0.3 K	24 Feb 2014 - 12:31	MattShotwell	example data
png	full-asymptotic-standard-error-comparison.png	manage	8.4 K	27 Jan 2014 - 11:58	MattShotwell
png	full-bootstrap-standard-error-comparison.png	manage	8.0 K	27 Jan 2014 - 11:58	MattShotwell
png	mean-absolute-error-comparison.png	manage	39.9 K	27 Jan 2014 - 10:58	MattShotwell
R	selection.R	manage	14.0 K	13 Jan 2014 - 12:45	MattShotwell	Code for simulating model selection strategies
png	simpsons-paradox.png	manage	24.4 K	27 Jan 2014 - 11:07	MattShotwell
R	stepwise.R	manage	4.8 K	15 Jan 2013 - 10:38	MattShotwell	Stepwise versus "Full Model" comparison script

Topic revision: r59 - 31 Mar 2014, MattShotwell

Main

Department Home Page

Biostatistics Graduate Program

Vanderbilt University Medical Center

Biostatistics Webs
- Archive
- Main
- Sandbox
- System

Copyright © 2013-2022 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding Vanderbilt Biostatistics Wiki? Send feedback