Welcome back. Get ready for more quantitative methods!!!
If you are looking for the VSS, they are here: VSS. The rest of the material for first term is at the bottom of this page. Some of it will get deleted during the term.
The syllabus for this term is here. Details on the grading are there. The order and content of lectures may change, so each week look at this web page.
How to for Spring 2012 here.
Journal to turn in week 1. Submit what you did for the virtual statistics sessions and any questions that came up while you were doing these. Don't take long on this, just write a quick summary.
Lecture 1. Outline, tea drinking, statisticians, maybe some signal detection theory, and matrices
Reading (*s are required)
* Read Fisher, chapter 2, if you did not during the break-from-all-the-boring-non-stats stuff.
Matrix handout that will be given to you (read if you don't know matrix algebra).
Wright, D.B. (2009). Ten statisticians and their impacts for psychologists. Perspectives on Psychological Science, 4, 587-597. http://www2.fiu.edu/~dwright/pdf/tenstats.pdf. This is discussed in the lecture, so use this rather than taking notes during that part of the lecture.
For next week
* Any introductory book on their introduction to ANOVA and ANCOVA (Field, 2009, Ch. 10; Wright & London, 2009, Ch. 7; etc.).
Journal. See slides.
Story of Nala and Kali here
Slides here Read here if using slides
Week 2 - January 18
ANOVA 1 /ANCOVA
Main reading
focusing on SPSS
* Field, A.P. (2009). Discovering statistics using SPSS (and sex and drugs and rock 'n' roll) (third edition). London: Sage publications. Chapter 11 (Analysis of covariance, ANCOVA). OR any chapter on ANCOVA
focusing on R
* Wright, D.B., & London, K. (2009). ANCOVA: Lord's paradox and mediation analysis. In D.B. Wright & K. London's Modern regression techniques using R (pp. 48-64). London: Sage. We go through the forgetting example from the lecture. This was distributed last week.
On Lord's paradox (all optional, but at some point at least read Lord's two page paper!)
Lord, F. M. (1967). A paradox in the interpretation of group comparisons. Psychological Bulletin, 72, 304-305. A brief and provocative paper from Lord.
Wainer, H. (1991). Adjusting for differential base rates: Lord’s paradox again. Psychological Bulletin, 109, 147-151. Wainer uses Rubin's model of causality to explain Lord's paradox. This is available through PSYCHArticles on the library website. I really like this paper as a description of Rubin's model.
Wright, D. B. (2006). Comparing groups in a before-after design: When t-test and ANCOVA produce different results. British Journal of Educational Psychology, 76, 663-675. Code. This is more computational than Wainer's paper, but shows that ANCOVA is usually preferred over differences.
London and Wright have a chapter, "Analyzing change between two or more groups: Analysis of variance versus analysis of covariance", in the Handbook of Developmental Research Methods. Those of you dealing with children might want to read this instead of the Wright and London from above. First of you to come by my office can borrow the Handbook. I think some of the chapters in this Handbook will be readings for the Developmental Methods course.
Lecture Slides Note on using slides
Week 3 - January 25
ANOVA 2 / Factorial
** The McGuire and London chapter emailed around
Field, A.P. (2009). Discovering statistics using SPSS (and sex and drugs and rock 'n' roll) (third edition). London: Sage publications. Chapter 12.
Wright, D. B. & London, K. (2009). First (and Second) Steps in Statistics (2nd ed). London: Sage. pp. 159-169.
Useful websites
http://psych.wisc.edu/moore/R_Analysis_of_Variance_Handouts_html.html
http://personality-project.org/r/#anova
Example from:
Berdoy, M., Webster, J.P., & Macdonald, D.W. (2000). Fatal attraction in rats infected with Toxoplasma gondii. Proceedings of the Royal Society of London, B, 267, 1591-1594. R code here.
Lecture Slides Note on using slides
Week 4 - February 1
Multiple regression
Lecture slides Note on using slides
Reading
Main reading:
Jeremy Miles' chapter was emailed.
The Wright and London chapter from MRT was sent around.
Other:
Bartholomew et al. Chapter 6, but not the parts on logistic regression (we do that next). We will have a few copies of this available.
On SPSS
Field Chapter 7
More intro
Wright and London, First Steps, the second half of chapter 9
Journal
Use the following data on how a rat's behavior may (or may not) be affected by Toxoplasma Gondii. Data. Make a graph that is good for describing the data.
Week 5 - February 8
Generalized linear regression (sung to the theme tune of "Attack of the Killer Tomatoes" ... not as cool as last week's film Easy Rider, but it has its place in cinema)
Reading:
Bartholomew et al. Chapter 6, the part on logistic regression.
Field, chapter 8
Wright and London (Modern Regression Techniques) Chapter 6
Lecture Slides Note on using slides
Week 6 - February 15
MANOVA / Discriminant function analysis / Multilevel
Field, A.P. (2009). Discovering statistics using SPSS (and sex and drugs and rock 'n' roll) (third edition). London: Sage publications. Chapter 16. This is great coverage of MANOVA, using discriminant analysis as a graphical method for it (it can be used in other ways).
Most multivariate books cover these topics (in more detail than Field), so if you want more, try, for example, Tabachnick and Fidell (here for her art, or here for the book). T&F is designed for psychologists, and is a great book, just costs more than I wanted you to spend.
The multilevel chapter in Wright and London (modern regression techniques), sent around. It focuses on R. Field has a multilevel chapter too. I think multilevel in SPSS is trickier than R.
The multilevel chapter in Bartholomew is also good. I'll bring that book to class. Bartholomew, Steele, Moustaki & Galbrath (2008). Analysis of multivariate social science data. Chapter 12. http://www.cmm.bris.ac.uk/team/amssd.shtml
If you are a dev person, consider getting:
Singer, J. D., & Willett, J. B. (2003). Applied longitudinal data analysis: Modeling change and event occurrence. Oxford University Press. Great book for using multilevel models with longitudinal data (see here). Examples for lots of stats packages, and Singer is working on a multilevel book.
Goldstein, H. (2003). Multilevel statistical methods (3rd edition). London: Edward Arnold. For statisticians, but well written. The second edition is freely available here. Page for the 3rd edition is here.
Chapter 1 of Goldstein (2003).
Lecture Slides Note on using slides
Some of the code I used is here, and some xxx.
Future lectures put up on the Monday before the lecture
&*(*&(*&(*&(*&^^^^^^^^^ FIRST TERM O*I&*(&(*&(*&(*&(*&(*&
SYLLABUS (but it gets updated! Check the web)
Course Structure
The typical lecture will be me talking for a while about the topic, having a 120
seconds for a coffee break, and then two or three of you presenting "how-to"
exercises. Sometimes we will have time to work through exercises and Andrea and
I will walk around to help with these.
Reading (* required, # worth doing). Many of the items listed include links to
the papers. Email Andrea or me if a link is not working. Where it is says
"slides", the up-to-date ones get put on the Monday of the lecture. Many of last
year's slides are still up, so you can look ahead, but the topics sometimes
change a bit (particularly if our timing gets blown off).
Many of these reading are hyperlinks to the sources.
Week 1 (Aug 24) . Introduction, why statistics, what are SPSS and R, and what
psychologists want from statistics.
Reading
# Chapter 1 of First Steps. Read all the exercises
* The following introduction to R:
Chapter
One PDF
*
Facets of R
(Chambers, 2009).
# Paul Meehl on philosophy of psychology
here
The Star Trek stuff (or R, I get them confused) is from
Chambers,
J.M. (2008).
Software for data analysis: Programming with R.
Springer. This book describes in detail the philosophy and mechanics about R
(and therefore S, S-Plus). It is more for people interested in software
than people for people interested in statistics (but if you are really into
statistics you should be interested in the software too).
Week
2 (Aug 31): Boxplots, quantiles, histogram: intro to probability.
Reading.
* Chapter 2 of First Steps. Read exercises.
* Gould, S.J. (1985). The median isn't the message.
Discover, 6 (June), 40-42.
here.
# The intro from Jeffrey's book (a copy will be brought to class) or the Howell
chapter.
# Reese, R.A. (2005). Boxplots. Significance,
2, 134-135.
here.
The Vox Populi paper
http://galton.org/essays/1900-1911/galton-1907-vox-populi.pdf
The Stigler reference is: Stigler, S.M. (1983). Who
discovered Bayes's theorem? The American
Statistician, 37,
290-296.
Week 3 (Sept 7). Means, standard deviation, skewness, transformations, making
bad graphs.
* Chapter 2 of Steps
* Look at
http://www.edwardtufte.com/tufte/
# Bland and Altman on transformations
here.
Martin Bland and Doug Altman have dozens of brief articles on statistical
methods in British Medical Journal (kind of like what Reese does in
Significance). These are good sources.
# Howard Wainer's "How to display data badly"
here.
# Wright, D. B. & Williams, S. (2003). Producing bad results sections. The Psychologist, 16, 646-648.
Handout (I will bring to class so you don't need to print)
Week 4 (Sept 14). Categorical data, bar charts, measurement, ranking
* Chapter 3 of Steps
* Lord, F. (1953). On the statistical treatment of football numbers. American Psychologist, 8, 750-751. reprinted here in chapter form, and from the APA here (this may require you being on a university computer).
* Look at http://www.r-tutor.com/elementary-statistics
# (but you should read sometime) Stevens, S.S. (1946). On the theory of scales
of measurement. Science,
103, 677-680.
here.
# Further reading on Lord's example from a recent issue of
J of Math Psych
here.
# Reese, R.A. (2007).
Bah! Bar charts.
Significance, 4, 41-44.
I will do group 1 this week for the 8 heads of 26 coin toss stuff. Group 2's (where the coin tosser kept going until 8 heads were achieved) will be next week.
Week 5 (Sept 21). Sampling and allocation
* Chapter 4 of Steps
*
Wright, D. B. (2006).
Causal and associative hypotheses in psychology: Examples from eyewitness
testimony research.
Psychology, Public Policy, and Law,
12, 190-213. This is a
long paper, but covers really the main points I want to make in the lecture.
* Feinberg, S. E.
(1971). Randomization and social affairs: The 1970 draft lottery.
Science,
171, 255-261. This is a great
description of what probability is!
#
Bland, M. (2005). The Horizon
homeopathic dilution experiment. Significance,
2, 106-109.
Should be able to access
here.
This is a good descriptions of the complexities of random allocation. The study
was done for the BBC. BBC coverage
here.
It began with a paper that supposedly showed homeopathy worked. Bland
describes how failures in randomization may have produced these results and
concludes about the original author of the Nature
paper: "I think it is safe to say that this episode destroyed Benveniste's
career" (p. 107). Significance
is the magazine for the Royal Statistical Society.
It is pretty cool!
# Cronbach, L. J. (1957).
The two disciplines of scientific psychology.
American
Psychologist, 12, 671 -684. This is an important paper in the history of
psychology. You should read this at some point in your PhD.
# Holland, P. W. (1986). Statistics
and causal inference. Journal
of American Statistical Association,
81, 945-960. This is a
fairly long paper. It describes Rubin's model of causality, which we will
discuss at a few points in the course (and which is described in the APA Task
force report). It is aimed at statisticians, but is not heavily
mathematical.
# Starr, N. (1997). Nonrandom Risk: The 1970 Draft Lottery.
Journal of Statistics Education,
5 (2).
http://www.amstat.org/publications/jse/v5n2/datasets.starr.html.
This is about the same data as Fienberg’s describing using it as a teaching
resource. It includes some of the data.
Also cited in lecture:
"Student" (1931).
The Lanarkshire milk experiment. Biometrika,
23, 398-406. This is a
good read. Critical commentaries were more direct back in the 1930s.
Week 6 (Sept 28). Inference, confidence intervals, power
* Chapter 5 of First Steps
*
Baguley, T. (2004). Understanding statistical power in the
context of applied research. Applied Ergonomics,
35, 73-80. I
really like this paper! It points out that people should think about why
they are doing power analysis.
* Cohen, J. (1988). A power primer. Psychological Bulletin, 112, 155-159. This paper provides tables that can be used to calculate power and this will be explained in a how-to session. It also describes what Cohen calls small, medium, and large effect sizes (but lots of people warn not to use these blindly).
Power tutorial by Jeremy Miles here. This is a nice tutorial for power!
To
download the G*Power program press
here.
For a tutorial on G*Power see all the links in the left column of their web page.
Lenth's
power page
here.
Week
7 (Oct 5). t-test, Wilcoxon, NHST and alternatives
Student's
t-test is covered pretty well in every intro textbook.
His 1908 paper was a real landmark (he used z rather than t, I have the paper
somewhere). Wilcoxon and Mann-Whitney is covered less (and used less). They are covered
them briefly in First Steps, but other intro books tend to cover them a bit more). The classic book for teaching ranked-based methods is Siegel
and Castellan, but it has a different approach than how people write now (i.e.,
they show how to do it by hand in a recipe like fashion).
* Chapter 6 of First Steps.
*
Cohen, J. (1990).
Things I have learned (so far). American
Psychologist, 45, 1304-1312.
A wonderful read about how to think about statistics.
Paper available through FIU library in journal format.
# Meehl, P. Several publications
here,
including the two cited in the lecture, published in
1967
and
1978.
Look at one of these two. The web page is also an
excellent source for taxometric analysis and how people make bad decisions.
# Cohen, J. (1994). The earth is round (p < .05).
American Psychologist, 49,
997-1003. More specific than his 1990 paper.
This paper focuses on NHST. This is on FIU e-library.
#
Wilkinson, L. and the Task Force on Statistical Inference, APA Board of
Scientific Affairs (1999).
Statistical methods in psychology journals: Guidelines and
explanations. American Psychologist, 54,
594-604. The APA report! You should have
already read this, but it is worth reading again.
Week 8 (Oct 12). ANOVA, Kruskal-Wallis, etc., and the Bootstrap
* Chapter 7 of First Steps
* The Wright part of: Wright, D.B. & Field, A.P. (2009).
Giving your data the bootstrap
(earlier version). Psychologist,
22, 412-413.
*
R Guide -- Analysis of Variance
ANOVA (particularly repeated measures) is tricky in R (for some good reasons).
# Richard Lowry on ANOVA
here.
The VassarStats site is good. One of his big projects is trying to get a
web-based free statistics software (with lots of how-to info).
# Efron, B & Gong, G. (1983). A Leisurely Look at the Bootstrap, the
Jackknife, and Cross-Validation. American Statistician, Vol. 37, No. 1,
(Feb., 1983), pp. 36-48.
here.
This is a stats introduction, so skim the equations.
Week 9 (Oct 19). Regression intro
* Chapter 8 of First Steps
* Handout
All these available from FIU electronically.
# Doherty, M.E., & Anderson, R.B. (2009). Variation in
scatterplot displays. Behavior Research Methods,
41, 55-60. The authors
talk about guidelines for making scatter plots, but most important, stress
thinking about how to make scatter plots.
# Reese, R.A. (2006). Scatterplots,
Significance,
3, 136-138.
# Reese, R.A. (2008). Scatterplots revisited.
Significance,
5, 87-89. [This one is on
adding more info into scatter plots but keeping 2-d.
Ocean temperature data:
fishstock.sav
(SPSS),
fishstock.dat
(in ASCII for read.table in R)
So, for .sav either click and open in SPSS, or save onto your computer and use
read.spss.
For .dat, do fishstock <- read.table("http://www.fiu.edu/~dwright/qm4psych/fishstock.dat")
Week 10 (Oct 26). Chi-square
* Chapter 10 of First Steps
* Get your R functions stuff ready.
Agresti, A. (2002).
Categorical data analysis (2nd Edition).
Hoboken, NJ: John Wiley & Sons. This is a textbook appropriate for an
entire course on categorical/qualitative data analysis. It is more
detailed than most of you will need (and goes beyond the content of this course
and the next one ... if I teach a course on CDA, this will be the text). Alan
Agresti has a web site for this book
here.
On correspondence analysis,
here
is some R code.
Week 11 (Nov 2)
Presenting your functions
Week 12 (Nov 9)
. Advanced Glance and review
This is a review lecture, and looking forward. You have enough work you don't
need more required reading.
Welcome to Quantitative Methods for Psychologists 1. This
is a course for psychology PhD students and if you are taking this you should
also be taking the Wednesday afternoon course with Dr. Stephens. If you plan to
audit it (which requires permission and the room may be near maximum occupancy),
you still have to do some assignments and presentations. The content of the
course is a rapid review of some underlying issues in statistics, review of
basic tests (like ANOVA, regression, Chi-square), and making sure that you can
conduct these analyses in SPSS and R. The emphasis is on knowing
why
you are doing the statistics. It is expected you already learned
how to
do most of these statistics during your undergraduate.
Teaching assistant:
Andrea Arndorfer (aarndorfer@gmail.com). She will discuss various issues about the course with you and
we will discuss when will be the best time for her office hour.
My Office hours will be sometime on Monday. We will discuss
when
during the first lecture to make sure it fits into your timetables. Email me (dwright@fiu.edu)
so I know when you are coming and if there are several of you.
Learning Outcomes
Applying basics statistics and graphical procedures to psychological research.
Requirements
40% journal (details below)
20% two how-to/crib sheet page to go along with your presentations
40% exam
0% group function (but you must get A- or better to
receive either a happy face or a star)
93% A
90%
A-
87% B+
83%
B
83% B-
77% C+
73%
C
73% C-
67% D+
63%
D
63% D-
below F
Below B for PhD track students may create problems. So, do well.
Attendance.
Required, but email Andrea and me
before the lecture
if you have to miss due to illness, religious reasons, conference, etc. If you
use facebook and things like that during lectures, I'll do what I can to get you
removed from FIU.
Cheating.
Don't. If you are planning on cheating see me beforehand to discuss the
penalties.
Journal (40%)
From Week 2 onwards you will have little assignments and exercises to do each
week. Bring these with you. Some weeks we will discuss them (sometimes handing
them to others in the class to evaluate). At the end of each week it should be
submitted to Andrea who will mark them. You must turn in all of these. Andrea
and I will discuss them and contact people who are not keeping up.
How-to pages (20%)
In groups you will write a brief handout (one page in whatever format you think
facilitates learning) for classmates to supplement your presentation. You have
at least two presentations. The handouts must be handed out to the class and an
electronic copy can be put on this web page. It is normal for us to meet
on the Monday to talk about the presentation. Put your names on the handouts.
There will often be a biography that you will need to do.
Group R function (0%)
In groups (of probably 3 or so) you will write a function to do something in R.
As a group you will present these.
Exam (40%)
This will allow you to show that you can interpret output from R and SPSS and
that you understand some statistical concepts.
Reading
Books
(if you are in a lab with others, see what they have. I wouldn't buy anything
until you see what others have)
Field, A.P. (2009).
Discovering statistics using SPSS: and sex and drugs and
rock 'n' roll (third edition). London: Sage
publications.
Amazon.
$60, down from $80. There are older versions of this around, but the new version
does a lot more. This course will focus more on the SPSS bit that the S, D and
R&R. Don't get the DVD version, most of that stuff is on his web page. An R
version should be out for second term.
Wright, D. B. & London, K. (2009).
First (and Second) Steps in Statistics (2nd ed).
London: Sage.
Amazon
$35 down from $46.
Software
Many of you will have already encountered SPSS. It is a fairly common package.
Most of you are TAs/RAs so as staff members you can get a discounted version.
Again, check around your lab.
The second software system is R, which is rapidly growing in academic circles
and beyond. There are about a 100 books on it, but also much that is freely
available (as is the program, it is free, see
http://cran.r-project.org/).
I think we can get by just using free sources for this course.
No need to buy a new calculator; computers all have these. As a PhD
student, you will want to have access to a computer at home.
Articles
Throughout this course there will several articles to read.
I will try to choose ones that can be accessed electronically through the FIU
library page. As part of doing a PhD you should read, throughout your PhD time
and beyond, the occasional methods paper. Niel Waller put together a good start
here http://www.psych.umn.edu/faculty/waller/readings.htm, and the "classics"
page has some too (see http://psychclassics.asu.edu/topic.htm#statsmeth).
Data
Most of the data used for this course is either
http://www.uk.sagepub.com/field3e/SPSSdata.htm for Field or
http://www2.fiu.edu/~dwright/1ststeps/index.htm for Wright and London.
Week 11 (Nov 2)
Presenting your functions
Week 12 (Nov 9) Applied Measurement
Week 13 (Nov 16). Advanced Glance and review
This is a review lecture, and looking forward. You have enough work you don't
need more required reading.
Week 14 (Nov 23)
EXAM
####Below is from last year, and some of it will be used after the break. Look at if you want, but some of the links aren't active.
This course is the second term of a two-term sequence in quantitative methods for PhD Psychology students. The first term's materials are down below. This site is just for the timetable for this term.
Week 3 - January 26
ANOVA 2 / Factorial
Field, A.P. (2009). Discovering statistics using SPSS (and sex and drugs and rock 'n' roll) (third edition). London: Sage publications. Chapter 12.
Wright, D. B. & London, K. (2009). First (and Second) Steps in Statistics (2nd ed). London: Sage. pp. 159-169.
Example from:
Berdoy, M., Webster, J.P., & Macdonald, D.W. (2000). Fatal attraction in rats infected with Toxoplasma gondii. Proceedings of the Royal Society of London, B, 267, 1591-1594. R code here.
Lecture Slides Note on using slides
Journal
Use the following data on how a rat's behavior may (or may not) be affected by Toxoplasma Gondii. Data. Make a graph that is good for describing the data.
Week 4 - February 2
MANOVA / Discriminant function analysis
Field, A.P. (2009). Discovering statistics using SPSS (and sex and drugs and rock 'n' roll) (third edition). London: Sage publications. Chapter 16.
Most multivariate books cover these topics (in more detail than Field), so if you want more, try, for example, Tabachnick and Fidell (here for her art, or here for the book). T&F is designed for psychologists, and is a great book, just costs more than I wanted you to spend.
Lecture Slides Note on using slides
Some of the code I used is here, and some here.
HOW TOs: I changed my mind. Rather than one group doing MANOVA without a group variable, both groups use the same dataset (or one you make yourselves) and one group do MANOVA and one group do discriminant function analysis. Field splits his chapter in this way. If you want a dataset, here are some data for a study currently running. The study is more involved, and has about 3 times more participants, but this is a good number for MANOVA illustration. There are three conditions. A confederate either is likeable or dislikeable, and there is a control group. After doing some memory tasks and before debriefing, participants are asked to make ratings about the confederate. These are in this file. They make lots of ratings. Choose 5 or 6 and use these in the MANOVA (with the group variable).
Journal
Either run MANOVA on the OCD data in R, or run MANOVA on the religion data in SPSS. The OCD data are here (all the book's data). The religion data are here and they come from www.thearda.com/Archive/Files/Downloads/BRS2005_DL2.asp. The ARDA (Association of Religion Data Archives) is a really good source.
Week 6 - February 16
Generalized linear regression
Reading:
Bartholomew et al. Chapter 6, the part on logistic regression.
Field, chapter 8
Wright and London (Modern Regression Techniques) Chapter 6 **** We will have copies of this
Lecture Slides Note on using slides
http://www.deathpenaltyinfo.org/murder-rates-nationally-and-state#MRord
Week 7 - February 23
Robust regression and Multilevel modeling
Full but a year old Robust lecture slides Note on using slides
Full but a year old Multilevel lecture slides
For robust:
Wilcox, R. R. (1998). How many discoveries have been lost by ignoring modern statistical methods? American Psychologist, 53, 300-314. But see also his books.
Wright, D.B. & London, K. (2009). MRT. Chapter 9 on Robust Regression. This will be handed out in class.
The exercise in class is shown here.
For multilevel
Best web site: http://www.cmm.bristol.ac.uk/
Readings (from easiest to more difficult)
Wright, D. B. (1998). Modelling clustered data in autobiographical memory research: The multilevel approach. Applied Cognitive Psychology, 12, 339-357. Intro focusing in multilevel ANOVA, regression, and logistic regression. A little old and pre-dates R and SPSS having multilevel capabilities.
Field, A.P. (2009). Chapter 19. Focusing on how-to with SPSS. SPSS' multilevel procedures are a little tricky to use. It is well suited for running longitudinal methods.
Bartholomew, Steele, Moustaki & Galbrath (2008). Analysis of multivariate social science data. Chapter 12. http://www.cmm.bris.ac.uk/team/amssd.shtml
Singer, J. D., & Willett, J. B. (2003). Applied longitudinal data analysis: Modeling change and event occurrence. Oxford University Press. Great book for using multilevel models with longitudinal data (see here). Examples for lots of stats packages, and Singer is working on a multilevel book.
Bartholomew, Steele, Moustaki & Galbrath (2008). Analysis of multivariate social science data. Chapter 12. Page is here.
Wright, D.B. & London, K. (2009). Multilevel modelling: Beyond the basic applications. British Journal of Mathematical and Statistical Psychology, 62, 439-456. This goes through both nlme and lmer. As the title suggests, it is the next step beyond this lecture.
Pinheiro, J. C., & Bates, D.M. (2000). Mixed-effects models in S and S-Plus. New York: Springer. Web page is here. The manual for nlme (Bates is working on one for lme4).
Goldstein, H. (2003). Multilevel statistical methods (3rd edition). London: Edward Arnold. For statisticians, but well written. The second edition is freely available here. Page for the 3rd edition is here.
Chapter 1 of Goldstein (2003).
Week 8, March something
Principal Component Analysis
Readings:
Wright and Villalba (in press) here, and web page here
Bartholomew et al. Chapter 5
Week 9
Exploratory Factor Analysis
Wright and Villalba (in press) here, and web page here
Bartholomew et al. Chapter 7
Week 10: Latent trait models
Item response modeling:
* Bartholomew et al. Chapter 8.
Rizipolous, D. (2009). http://cran.r-project.org/web/packages/ltm/ltm.pdf This is the software we will use.
Embretson, S. E. & Reise, S. P. (2000). Item response theory for psychologists. Mahmah, NJ: Lawrence Erlbaum Associates.
* Reise, S.P., Ainsworth, A.T., Haviland, MG. (2005). Item response theory. CD in PS. Great intro available on psychologicalscience.org. Henard, D. H. (2000). Item response theory. In L. G. Grimm & P. R. Yarnold (Eds.) Reading and Understanding More Multivariate Statistics (pp. 67-97). Washington, DC: American Psychological Association. This is a good introductory chapter. A previous student "borrowed" this book from me and I didn't get the book back.
* Zickar, M.J. (1998). Modeling item-level data with item response theory. Current Directions in Psychology, 7, 104-109. This is a review article aimed at postgraduates. It is available from psychologicalscience.org.
Wright, D. B., & Skagerberg, E. M. (2006). A dialogue about MCQs, reliability, and item response modelling. Psychology Teaching Review, 12, 66-79. The link is to the whole issue, just print what you want.
More stuff from Niel Waller's page here.
Latent class modeling:
* Bartholomew et al. Chapter 10.
Seminar at UCLA here (using MPlus)
Brief introduction (but some math) here
Brief introduction (bit less math) here
There are other R packages and non-R packages for latent class modeling. We are using the simplest I could find.
For intro book on taxometrics, see http://www.tcnj.edu/~ruscio/taxometrics.html.
March 30: A Statistical Tapas
I will add some reading for each of these, but there will not be a journal entry associated with this lecture nor will it be on the exam.
Bayesian slides
The main package for Bayesian analysis is the BUGS family (BUGS page), which stands for Bayesian inference Using Gibbs Sampling. Gibbs sampling is a computer intensive estimation technique that has proved to be really good. A good introduction to Bayesian inferences, comparing it with the traditional approach, is Dienes (in press) here.
Neural Nets slides
There is a large literature on neural nets as models of cognition. There they are usually considered as a distributed model with lots of nodes and connections. In biology sometimes people try to take into account the complexities of the individual neurons. The statistical procedures are very different. They are more of a black box method. They are a way to see if there is someway to connect and input pattern with an output pattern. Ripley describes alternatives for this which are better, and I believe him. The MASS book by Venables and Ripley provides a brief introduction to the topic and the function they wrote, and the Hastie et al. Elements book also provides a brief treatment. Both of these require more math than is assumed in this course.
Missing values slides
In the mini-lecture I will talk about statistical methods for dealing with missing data. Much more valuable though are the methodological means of reducing them. Andrea will bring some copies of chapter relevant to this.
Cluster analysis kmeans
There are lots of statistical methods for placing people and items into clusters. During this course we have already seen discriminant analysis, latent class analysis, and if you went through all the slides from last week taxometric analysis. There are lots of others. The phrase 'cluster analysis' usually refers to two general approaches. One is hierarchical clustering which provides a picture of how either the items of the people can be clustered. You define a dissimilarity measure and then the clustering algorithm puts similar items/people together. It either starts with every item/person in one big cluster and begins splitting the group up, or it begins with each item/person separately and begins grouping items/people. You get a nice map, but it is highly dependent on how the grouping is done, so it is worth playing around with many of the techniques in you software. The chapter in Bartholomew et al provides a good introduction at a level appropriate for this course. The other approach is to begin with the number of clusters that you want. It is called K-means or K-medoids. Hastie et al describe how to choose the number of clusters using the gap statistic (paper here), but it is dependent on why you are doing the clustering.
************************************************************************
Last year's here