The Analysis Institute - Making Statistics Make Sense

Online Workshops

Assumptions of the General Linear Model and How to Check Them

Overview:

Did you know that there is no assumption in linear regression and ANOVA that the dependent variable follow a normal distribution?

Really. The assumption is about the residuals.

In fact, while it doesn't happen often, it can happen that a non-normal dependent variable--often a bimodal variable--still meets the assumptions of the model.

I have talked with many researchers who are checking the wrong assumptions of the GLM. Or they're checking the right assumptions, but with over-conservative tests that find every data set fails to meet assumptions.

It's not surprising. Read any two statistics textbooks and they'll have a different list of assumptions. And list different ways to test them.

It's because while some of the assumptions are explicit, others are just implicit in the way the model is conceptualized. And some data issues, such as multicollinearity, while important to consider, are not actually assumptions of the model.

In this workshop, we will investigate each of the assumptions of the GLM so they make sense. You'll see they really aren't just a list of arbitrary rules.

Then we'll show how to check each one using plots, rules of thumb, and tests. We'll discuss which tests are too sensitive and when, and how to tell if a plot is "close enough".

And finally, we'll discuss, in detail, what your options are for dealing with problems and when and how to do them.

What You Will Learn:

Module 1: The Assumptions and Checking them

You will get an accurate understanding of each of the assumptions, both implicit and explicit. In just two hours, you will learn:

A brief review of the General Linear Model, in terms of regression and ANOVA, and how it directly and indirectly leads to the assumptions
what each assumption really means and why each assumption is important to maintain the integrity of the model and the accuracy of the p-values
how to check each assumption, some with tests, and some with plots
how to construct and read residual plots--how "robust" of a departure from normality or constant variance is acceptable
how to construct and read normal probability plots, which are much more accurate than histograms for diagnosing normality
common distributions and situations in which assumptions will likely fail, including count and categorical outcomes, and trucated and censored outcome data

Module 2: Your Strategies and Options If Assumptions Fail

what to do if one or more assumptions are not met, including transformations, weighted least squares, and modifying the model
All about transformations--the different types, why and how to do them, when to do them, on which variables to do them, and what they mean
Box-Cox transformations for determining whether and which transformation will best suit your data
A detailed, step-by-step example of how to run a model and check all the assumptions on a real data set (not a perfect textbook example

Format:

Web based online workshop: Each week, we'll release one module, so you don't get overwhelmed with too much information all at once. For each module, you'll get:

1. Training Materials

training videos to download and watch on your own time (and keep forever). The videos are screen shots of what is happening on my computer screen. They include both slide presentations as I go over the concepts and software demonstrations.
pdf handouts of the presentation slides for note-taking
exercises to practice what you've learned, with answers to check your work.

2. A Live Q&A Webinar

After you've had about a week to work through all the training materials, we'll meet in a live Question & Answer webinar.

You attend over the internet. Audio is through either your computer speakers/microphone or by telephone. Ask questions out loud or write it into the chat. You'll be able to see my screen, so I can show you how to do something in SPSS, draw you a graph, or we can review the output.

I highly encourage you to attend the Q&A Sessions live, but if you have to miss one (it happens) they will be recorded and made available within 24 hours on our workshop website. You can still get your questions answered by submitting a question before a Q&A and I'll cover it in the session.

3. Private workshop website where you can submit questions and get all the workshop materials.

This private website contains everything you need: Workshop materials to download, bonus videos, the exercises, Q&A information and recordings, a contact form to get some help if you get stuck with anything.

But best of all, on each Module page, you can submit questions. So if something isn't clear while you're watching the training, pause your video and type in your question. We'll answer quick questions right there on the website, and things that are more in-depth, we'll cover in the next Q&A.

It's all the advantages of a live workshop without the disadvantages.

You don't have to drop a few thousand travelling to a workshop and take a few days off work. And best of all, because you’re not travelling, we don’t have to concentrate an overwhelming amount of information into a marathon session. We can spread it out in digestible amounts.

So you really learn it. Inside out.

And best of all, unlike live workshops, you keep the workshop recordings forever. So if you need to review the material in a few months, or years, or find yourself needing to run a new type of model, you can watch it again.

Who Is It For:

This workshop is for you if you:

Will need to implement either linear regression or ANOVA soon, and want to be sure you're checking assumptions correctly
Have conducted a linear model in the past and are confused by the different things you've heard about its assumptions and how to check them
Are about to defend your dissertation and want to be sure you thoroughly understand your analyses well enough to defend your decisions to your committee
Would like to learn as much as you can about regression and ANOVA, so that you can apply it, and learn its extensions, with confidence and real understanding
Are comfortable with what the assumptions are, but would like to see, for once, how to really interpret the graphs or run transformations with confidence that you're doing it correctly

Prerequisites

Although we will use SPSS to show the examples, the focus here is not on the software. This workshop will be useful to anyone using any software. That said, you should be familiar with running an ANOVA or linear regression in some statistical software. The data sets will be available in SPSS, SAS, and excel format so you can import it into the statistical software of your choice.
You will get the most out of the workshop if you have had at least two statistics classes and some experience in data analysis.

What's Included:

2 Modules of training videos. Each module is 1.5-2 hours of training.
2 60-minute Webinar Q&A sessions. Just your questions and my answers. These are about a week after the training materials are released so you have time to watch the videos, do the exercises, and even try things out on your own data. Then come back with questions. It's when you use these techniques on your own data that you cement your learning.

PDF handouts of the presentation, which will be available ahead of time, on which you can take notes.
Data files and SPSS and SAS code to run, explore, and practice all of the workshop examples yourself.
Access to our private workshop web site. Submit questions, get call-in information for Q&A sessions, ask for customer support, and download handouts, recordings, and bonus materials.
You are encouraged to ask questions about implementing these techniques in your own analyses in addition to general questions about the topic. (But please keep questions to the topic of the workshop so that they’re useful for everyone).
Exercises and answers. (Yes, homework!). You’ll get more out of the workshop if you participate fully by doing homework. We suggest you use one of the data sets provided, so you can check answers, then redo them on your own. It's when you do statistics on your own data that it really starts to make sense.

Have a question about whether you're doing it right on your own data? Bring it to the Q&A session.

The Instructor:

Karen Grace-Martin is a statistical trainer and consultant and an expert on linear models, SPSS, and SAS.

She has guided and trained researchers through their statistical analysis for over 15 years. Her focus is on helping researchers gain an intuitive understanding of how statistics is applied to real data in research studies.

What our Students Say About Our Workshops:

"The examples were easy to understand and including demonstrations in the presentations were very helpful. Having the data sets to play with ourselves in addition to the specific homework exercises really helped to make this a major learning experience. Karen's slides are so clear and well organized, and she keeps just the right pace. I did not realize online courses could be so effective. Karen is an extremely effective teacher and maximizes use of technology to make this an extremely valuable learning experience."

-Prue Plummer-D'Amato, PhD

"I would definitely recommend the workshop to colleagues. It is one of the easiest and most friendly ways of explaining regression to non-statisticians that I have ever followed. Very useful."

-David Arteta, PhD

"The homework followed by live Q&A were were an excellent supplement to the videos. The materials and videos are so clear.

I was concerned that I would miss the sessions live, so to find out this time that all the workshops were pre-recorded and only the Q&A sessions were live was a relief. Having participated in previous workshops, I think this format is an excellent idea. Knowing I had to do the homework before the Q&A session made me make time to watch the videos and do the exercises every week instead of putting it off!"

-Anonymous

"The ability to hang on to all of these materials, including the bonus materials is a really great feature. I feel llike I have this whole library of references that I can go back to if I need it."

- Noelle Chesley, PhD

Dates and Times:

The modules will be released on:

Thursday, July 14
Tuesday, July 21

Question & Answer series:

Thursdays @ 12pm (noon) EST (GMT-5)

7/21/11
7/28/11

Wednesdays @ 8pm EST (GMT-5)

7/20/11
7/27/11

Registration:

PAY IN FULL: $147

If you would like a written, searchable resource to refer to after the workshop, upgrade to pre-order the full written pdf transcripts of the workshop for an additional $25.

WORKSHOP + Transcripts
PAY IN FULL: $172

Extras

If you would also like a written, searchable resource to refer to after the workshop, in addition to the videos, upgrade to pre-order the full written pdf ehanced transcripts of the workshop for an additional $25.

We call these enhanced transcripts, because not only do we have them transcribed by a real person and edit them extensively to make them easier to read, each import the presentation slides and screenshots so that you can see all the relevant graphs and equations as you're reading.

Free Bonuses:

To fill in your understanding about General Linear Models and where the assumptions fit into them, you'll get videos of:

A Review of Logarithms for the Data Analyst
The 11 Steps to Performing any Statistical Model
The First 3 Steps in Running any Statistical Model: Define and Design

Guarantee

Your registration fee is fully refundable up to 72 hours in advance minus a $25 administration fee. Because you will still receive the recorded video, no refunds will be granted after the program begins.

That said, your satisfaction is guaranteed. If you participate in the full workshop and find you are not satisfied for any reason, we will give you a full refund. Just notify us within 90 days of the conclusion of the program.