Supplemental Material

for Statistical Methods for Psychology, 9th ed.

This page lists supplemental material that is available for Statistical Methods for Psychology, 9^th edition. The material consists of a several pages about using the R computing environment and discussion of many separate topics that did not fit in the book itself. The latter is organized by chapter, but there is some material that does not fit neatly into a specific chapter. So hunt around.

Materials related to R

• Material on obtaining and running R


• A set of R programs. In the process of revising this book, I have created a huge number of programs in R. They have been compressed into a zip file, which you can download here. They are of varying states of quality, so I am sure that they are not all worth having. But I don't have the time to go back and review so many files. They are available for what they are worth--with no quarantee.

Supplements by Chapter

• Chapter One — Basic Concepts

• No supplemental material

• Chapter Two — Describing and Exploring Data

• Kernel-Density
A short description of how to calculate kernel density plots in R and SPSS.

• Chapter Three — The Normal Distribution

• Normal Probability Calculator
One of the best online calculators I know for computations for probability distributions. Goes way beyond the normal distribution.
• Normal Probability Calculator for the iPhone
This is a great calculator for the iPhone. You can get the free version (Statsmate Lite) or you can splurge and pay $0.99 for the full version. This calculator carries out tons of analyses--it's not just a probability calculator.
• More about Q-Q plots
  Murdoch University in Australia has a nice web page that helps you learn more about data by looking at the Q-Q plot.

• Chapter Four — Sampling Distributions and Hypothesis Testing

• Jones and Tukey Table
Andrew Waters, of Uniformed Services University of the Health Sciences, has suggested that a nice way to clarify the suggestion by Jones and Tukey is to present the decision process as in the attached table.

• Chapter Five — Basic Concepts of Probability

• Binomial Distribution Calculator
This link will allow you to enter the number of trials and the probability of success on any one trial, and calculate the probability of a specified number of success on those trials. You can also use either of the calculators referred to above.
• Multinomial Distribution Calculator
This link will allow you to enter the number of different possible outcomes and the probability of success for each of those, and calculate the probability of a specified number of success on those trials. For example, with 3 outcomes the probabilities might be p(black) = .50, p(green) = .20, p(pink) = .30. We can now find the probability of 3 blacks, 2 greens, and 5 pinks.

• Chapter Six — Categorical Data and Chi-Square

• Alternative Research Designs
This is a short entry that I wrote for the International Handbook of Statistical Sciences (2011), edited by Miodrag Lovric. It discusses the fact that we can construct contingency tables from a variety of different research designs, and these tables can be treated differently.
• Ordinal Categorical Variables
The standard chi-square test on a contingency tables is not affected by the order of the categorical variable. But when the categorical variable you are using is ordinal, you can obtain significantly greater power for your test by taking that ordering into account.
• Testing Change Over Two Measurements in Two Independent Groups
This entry addresses a question about two independent proportions of change. In other words, if Group A improved by 30% and Group B improved by 35%, is that difference significant. It started out as a simple question but soon became more interesting. I admit that I got a bit carried away.

• Chapter Seven — Hypothesis Tests Applied to Means

• Sampling Distribution of the Mean
A program written in R to demonstrate the sampling distribution of the mean for various sample sizes.
• Confidence Intervals on Effect Size
This entry should perhaps fit better under Chapter 18 because it deals with randomization tests. However it focuses on one- and two-sample tests, which is why it is here.

• Chapter Eight — Power

• Another Power Calculator
Russ Lenth at the University of Iowa has created a very nice free power calculator. It does not have all the bells and whistles of G*Power, but it is quite useful. One trick is to notice that there are little square boxes on each slider that allow you to change the range.
• What About Power When we are Concerned with a Meaningful Difference.
Several years ago I wrote a document on power when our null is not zero. We may want to reject the null only when the difference is sufficiently meaningful. This document raises important questions, though it does not attempt to answer all of them.

• Chapter Nine — Correlation and Regression

• Sampling distribution of the regression coefficient
This is another answer to a question you probably never expected to ask. If we are going to use a t test on a regression coefficient (b), then we would hope that b is normally distributed. But is it?
• Useful Source for Regression and Correlation
I have a Web page that contains a variety of examples on correlation and regression along with some other useful information. It may prove helpful for those seeking more information.

• Chapter Ten — Alternative Correlational Techniques

• Intraclass Correlation Coefficients
This is a discussion that I wrote in the past about the intraclass correlation coefficient. It might ask something that you don't care to know, but it gives an important perspective. It bases much of the discussion on an understanding of the analysis of variance, which is not covered until the next chapter, but I put it here because that is where people are likely to look for it.
• Kinds of ICC coefficients
The intraclass correlation is more complex that it might seem, depending mainly on how the data are collected. This document explains the various types. I believe that I wrote this, and a search of the Internet doesn't show that someone else did, but if I have taken someone else's work I apologize and would be happy to add an erratum.
• Reliability
In looking again to see if I could find out if someone else wrote the previous entry, I came across an excellent page on reliability. No author is given, but it is on the Hanover College web site, so it was probably written by John Krantz.

• Chapter Eleven — Simple Analysis of Variance

• Testing for homogeneity of variance using O'Brien's test
Tests of homogeneity of variance are often criticized, and the literature is full of papers showing that one or another of them is pretty poor. Hervé Abdi has a very good discussion of O'Brien's test on his web site.

• Chapter Twelve — Multiple Comparisons Among Treatment Means

• More on Multiple comparisons
In the text I said that I would provide a copy of an older chapter that covers a wider variety of multiple comparison techniques in greater detail. If you really want that stuff, here it is. Much of the chapter is similar to the one in the 8th edition, but there is a lot that has been left out of the new version.
• Another View of the False Discovery Rate
I could not afford the space to spend a lot of time on the False Discovery Rate in this chapter, but a more extensive discussion is available in a set of lecture notes by Bruce Walsh. I think that it nicely expands on what I said, but I still prefer the refernce to Maxwell and Delaney(2004).

• Chapter Thirteen — Factorial Analysis of Variance

• Unequal cell sizes
Unequal cell sizes can often make a significant difference in a factorial analysis of variance. There are alternative ways of treating unequal ns (see Type I II III.pdf), but the standard default, while generally appropriate, can sometimes give you misleading results.
• SAS for Random Effects and Nested Designs
In Section 13.8 of Chapter 13 I discuss how to deal with random effects and how to deal with nested designs. The two often go together. I have compiled SAS code for those analyses, and have presented the printout. The printout can be found at SAS for Alternative Designs and contains the links to the code and the raw data. You should know that there are other ways to write some of this code, but this will give you what you need. I would like to thank Karl Wuensch for his suggestions and for pointing out that the code that I give in the text, while it will work, is not optimal.

• Chapter Fourteen — Repeated-Measures Designs

• Multiple comparisons with repeated measures
• Analysis of Repeated Measures as a Mixed Model--Part 1
• Analysis of Repeated Measures as a Mixed Model--Part 2
While the standard repeated measures analysis of variance is one of the easiest ways of analyzing such data, the approach using mixed models has a great deal to recommend it, especially with missing data. Part II is less useful, and less well written, than Part 1, but I included it for completeness.
• http://www.bodowinter.com/tutorial/bw_LME_tutorial2.pdf
  The above two web pages are excellent as a way of addressing mixed models. I think that the second is probably more relevant.
• A Quickie on Trend Analysis
This is a quick document that needs a lot of work. It focuses on trend analysis as a way to draw out information on differences across a within subject variable.

• Chapter Fifteen — Multiple Regression

• Generating Data With a Specific Pattern of Correlations
We sometimes want to generate a set of data that has a particular pattern of correlations. These correlations can either be exact in the sample, or they can refer to populations having those correlations and the data are simply a sample from those populations.
• Another Example of Multiple Regression
It is always useful to look at multiple examples of a type of analysis, and the link here, from the Journal of Tropical Pediatrics--author unknown, contains an interesting and useful one related to crime rate.
• Multinomial Logistic Regression
In Chapter 15 we considered logistic regression with only two possible levels of the dependent variable. That is the normal form of logistic regression. However suppose that you had three outcomes--"improved","no change", "worse." This is a case of what is called multinomial logistic regression. An example is given at the site linked above.
• Mediation and Moderation
Some of the most important work on mediation and moderation has been done by David Kenny. These two links point to valuable pages that he has assembled on these topics.

• Chapter Sixteen — Analyses of Variance and Covariance as General Linear Models

• Another Example of the Analysis of Covariance
Germán Rodriguez at Princeton has put together an example of an analysis of covariance that you might find helpful. It doesn't say much more than the present text, but every time you read something written by someone else you have the chance of expanding your understanding.
• SPSS and the General Linear Model
UCLA's Academic Technology Services maintains an excellent set of Web pages on a wide variety of statistical topics. The one on using SPSS to solve the General Linear Model is particularly good and will be helpful for those working with SPSS. It goes beyond what I have been able to cover in the text.

• Chapter Seventeen — Meta-Analysis and Single Case Designs

• A Broader View of Meta-Analysis
Jamie DeCoster, at the University of Alabama, has compiled an excellent page on meta-analysis. His coverage differs from the one in this book because he spends more time in the beginning on discussing how to conduct, as opposed to how to analyze, a meta-analysis.
• What Does a Meta-Analysis Look Like
In recent years online learning has become an important tool in education. This study is a complete meta-analysis looking at the effectiveness of online learning. I include it not for its statistical aspects, but to give students an idea of what a meta-analysis paper is all about.
• Learn More About Single-Case Designs
In the text I only had the space to discuss simple A-B designs. But there are a number of other kinds of designs that over differing advantages and serve different purposes. This link is good at presenting a discussion of the strengths and weaknesses of alternative experimental designs.
• John Crawford's Page on Single-Subject Designs
John Crawford, at the University of Aberdeen, has published extensively on single-case designs. His web page is excellent and I recommend it highly. I have published a number of papers with John, but that really means that he does all the work and I tell him how great a job he has done.

• Chapter Eighteen — Resampling and Nonparametric Approaches to Data

• More (Much More) on Resampling
It should be apparent from reading the text that I am a strong supporter of resampling procedures. Several years ago I put together an extensive set of pages on resampling--both bootstrapping and randomization tests. These pages have the advantage and disadvantage that I was aiming at looking at broader issues of hypothesis testing rather than simply at discussing how to perform the tests. I think that what I have here is more thorough that standard coverage.