CALIFORNIA STATE UNIVERSITY, LONG BEACH
Geography 400-001 and -002:
Geographical Analysis (Multivariate Statistics)
Spring 2012 (ticket # 7602 / 7603 MW 2:00 - 4:20 p.m., LA5-352)
Instructor Information:
- Instructor: Dr. C.M. Rodrigue
- E-mail Address: rodrigue@csulb.edu
- Home Page: https://home.csulb.edu/~rodrigue/
- Telephones: (526) 985-4895 or -8432
- Office: LA4 103W
- Mailbox: LA4 106
- Office Hours: MWTh 1-1:50 p.m., and by appointment
Course Description:
Prerequisites: GEOG 200, or MATH 180, or equivalent. Examination of advanced quantitative techniques commonly employed by geographers in analysis of spatial phenomena. Topics to be covered include: multivariate statistical methods as models for geographical analysis. Emphasis on the applications of these techniques in geographical research, including the use of computers where appropriate.
Course Objectives:
- Understanding of the scientific method in geography and closely related disciplines
- Competence in research design
- Skill in advanced statistical and spatial analytic methods for handling bivariate and multivariate data
- Ability to interpret the results of these methods correctly
- Proficiency in the use of statistical software, spreadsheets, and Internet functions
Required Course Materials:
- Text: Grimm and Yarnold, Reading and Understanding Multivariate Statistics.
Recommended Course Reference Materials:
- Online text by StatSoft to support their Statistica software package, Electronic Statistics Textbook
- Craig A. Wendorf's manuals for univariate and multivariate statistics, http://www4.uwsp.edu/psych/cw/statistics/
- David M. Lane's online basic statistics text: Lane, HyperStat Online
- And, if your math has gone the way of the dinosaur, an online encyclopædia: Weisstein, World of Mathematics
- Calculator with bivariate statistical functions (TI-36X or comparable)
Open Source Statistical Software Resources:
- PAST, a full-featured multivariate stats package developed by O. Hammer, D.A. T. Harper, and P.D. Ryan http://folk.uio.no/ohammer/past/
- G*Power, a program for pre-study analysis of how big a sample you'll need for given significance and power requirements or for figuring out the power of a study given the sample size and significance level http://www.psycho.uni- duesseldorf.de/aap/projects/gpower/
- CrimeStat III, a spatial statistics program developed by Ned Levine with a National Institute of Justice grant for use in analyzing crime patterns in space but much more widely applicable http://www.icpsr.umich.edu/CrimeStat/
- WinIDAMS, a statistical analysis package capable of multivariate analyses, which was developed for free distribution by UNESCO http://portal.unesco.org/ci/en/ev.php-URL_ID=2070&URL_DO=DO_TOPIC&URL_SECTION=201.html
- OpenStat, a general and extensive stats package, including a downloadable stats textbook, developed by Bill Miller http://www.statprograms4u.com/
- PSPP, a new and still developing open-source SPSS work-alike, which can handle really huge databases (as in billions of records and variables!) http://www.gnu.org/software/pspp/
- OpenOffice, which includes a spreadsheet, word processor, viewgraphs program, database managment system, math handler, and graphics package. http://www.openoffice.org/
- LibreOffice, which has forked off OpenOffice because of concerns about the new owner of Sun Microsystems, which supported the open-source development of OpenOffice (the then-new owner, Oracle, donated the OpenOffice project to Apache Software Foundation, an open-source developer http://www.libreoffice.org/
- NeoOffice is another fork of OpenOffice, this one for Mac users: http://www.neooffice.org/neojava/en/index.php
- Quantum GIS is an open-source GIS package: http://qgis.org/
Grading:
I grade on a modified curve, based on several lab projects and a final project of your own design, each of them weighted the same (100 points each). The projects start out with a lot of coaching and opportunities to do "lab redemptions." As the semester goes on, you will be given more autonomy and responsibility for research design and execution, culminating in a final project in which you find a large database related to your own interests, formulate a hypothesis or a question for your data, design an appropriate methodology, and then interpret your results.Attendance is crucial in a course of this nature: If you miss something at one point in time, it can easily prevent you from understanding subsequent concepts. As a result, I will sporadically take attendance and use that record to decide among students with similar overall scores who gets the higher, or lower, grade.
Makeup Policy
Makeups are possible in the event of a documented unexpected emergency in a student's life or through prior arrangement with the instructor when the student has advance knowledge of a compelling conflict in schedule, including religious obligations and observances. Makeups under these two circumstances will not be penalized. All other makeup requests are subject to denial or serious penalty. This includes consideration for your record of continuous attendance.
Tentative Course Outline:
- Introduction
- Advanced statistics and geography and related disciplines
Software
- Bivariate Methods
- (Re)Introduction to Correlation and Regression
Simple linear correlation and regression
Algebraic invariants and transformations
Logistic regression
- Multiple Regression and Partial Correlation
- Creating models predicting one dependent variable
Picking which variables and how many to put into a model
Interpreting results
- Principal Components and Factor Analysis
- The idea of creating "artificial" factors
Pruning a blizzard of variables down (PCA and EFA)
Rotating eigenvectors to create the cleanest model
Testing models (CFA)- Detrended Correspondence Analysis
- Like PCA/FA, creates "artificial" factors
Data reduction method like PCA/FA
Works with data that don't conform to PCA/FA expectations
Especially good for ordination or zonation- K-Means Clustering
- Works in the p-dimensional space that PCA/FA does
Specify a number of desired clusters
Iterative assignment of cases to the "nearest" cluster
- Other Methods if we have time
- Student project
- Students select multivariate datasets of interest to them
Students choose appropriate methods to process their data
Students interpret their own results
University Withdrawal Policy:
It is the student's responsibility to withdraw from classes. Instructors have no obligation to withdraw students who do not attend classes and may choose not to do so. Withdrawal from a course after the first two weeks of instruction requires the signature of the instructor and department chair, and is permissible only for serious and compelling reasons. During the final three weeks of instruction, withdrawals are not permitted except in cases such as accident or serious illness where the circumstances causing withdrawal are clearly beyond the student's control and the assignment of an incomplete is not practical. Ordinarily, withdrawals in this category involve total withdrawal from the university. The College of Liberal Arts adheres to this policy strictly, and does NOT sign withdrawal forms in the final three weeks of class for other reasons.
Accessibility:
It is the student's responsibility to let me know at the beginning of the semester if s/he has a disability that will require accommodation. I am personally committed to making my classes accessible and providing accommodations that will help everyone compete on an even keel. I need to know about the issue at the beginning of the semester, though, so that we can work out a mutually reasonable and satisfying accommodation.Related to accessibility, this course will be set up on BeachBoard to enable convenient contact, though the core of the class's online content is this web page. You will need to have a CSULB e-mail account to use BeachBoard, however. Announcements and messages from me to the class may come by e-mail. If you do not check your CSULB e-mail account regularly, but use another account instead, please set your CSULB account so that it will forward messages to your other account. The CSULB Technology Help Desk is now available for students, by the way. The URL for the Help Desk is http://helpdesk.csulb.edu. Their telephone number is (562) 985-4959.
Policy on Cheating and Plagiarism:
Work that you hand in is assumed to be original unless your source material is documented appropriately. Using the ideas or words of another person, even a peer, or a web site, as if it were your own, is plagiarism. Cheating and plagiarism are very serious academic offenses. Students should brush up their understanding by consulting the discussion of cheating and plagiarism in the CSULB catalogue. The Geoscience Diversity Enhancement Project student workshops on scientific ethics contain a section on plagiarism that may be a useful review: https://cla.csulb.edu/departments/geography/gdep/ethics.html. Furthermore, students should be aware that faculty members have a range of academic actions available to them in cases of cheating and plagiarism, including failing a student on that particular work, failing a student in the course, and referring the case to Judicial Affairs.That said, I do encourage students to help one another out on the lab projects in this class. You learn best when you try to teach someone else, and it helps to hear the material from someone else besides the professor. If you have any questions about whether collaboration on lab problems is shading into plagiarism or cheating, please talk to me about your concerns. You can also indicate who you worked with on a given lab, just for insurance.
GEOG 400 Home | Dr. Rodrigue's Home | Geography Home | ES&P Home | EMER Home | CSULB Home | Scientific Calculator
This document is maintained by Dr. Rodrigue
First placed on the web: 01/29/01
Last updated: 01/22/12