Dealing with Discrete Count Variables
Once you learn the ins and outs of linear models, it can seem that you’re ready to tackle any dependent variable.
Not all numerical dependent variables are created equal! Some are discrete, not continuous:
-- Number of surviving offspring
-- Number of crimes committed in an area
-- Number of days in the hospital
It will seem like you’ve performed the right analysis, but your results will be inconsistent and biased. Fortunately, there’s a solution…
My goal is that by the end of the workshop, you’ll know how to spot when you need a count model and how to choose the most appropriate one, implement it, and understand the results. You’ll learn six different types of count models (plus all their varieties) and how to select the one that will work best for your specific data set.
In this six-module live, online workshop, we’ll cover:
-- What counts are and why they don’t work well in linear models
-- When a model is appropriate and how to evaluate a model
-- Working with Offset/Exposure Variables
-- Understanding Incidence Rate Ratios and the log link function
-- Steps to Running a Poisson Model
-- Interpreting Coefficients, Marginal Effects, and Interactions
-- Overdispersion & Negative Binomial Models
-- Model Fit, Assumptions, and Influence
-- Zero Truncated, Hurdle, Zero Inflated Poisson and Negative Binomial Models
This workshop is suitable for graduate students and research professionals in fields that use statistics. It’s for you if you:
-- Have tried to do a Poisson or negative binomial regression before, but found it confusing or difficult
-- Are comfortable running and interpreting linear regression or ANOVA and want to take your statistical skills to the next step
-- Know you will need to implement a count model soon
It’s not for you if you:
-- Are a statistics beginner and have never done a linear regression
-- Have an advanced degree in statistics and want theoretical knowledge of count models
This course is a 6-module live, interactive online workshop from Tuesday, July 16th, 2019.
During each webinar session, the instructor will cover core concepts, and will leave plenty of time to ask your own questions.
In addition to the six module sessions, you’ll also meet with your instructor for four additional Q&A sessions where you can get additional assistance on workshop concepts, deepen your knowledge, and clarify any questions.
As a participant in the Analyzing Count Data workshop, you’ll have access to a participant-only website, your workshop “hub.” That’s where you’ll access all workshop resources and material, including:
-- Real research data sets in SPSS, SAS, Stata, and csv formats.
-- Syntax and demonstration videos to conduct all the workshop examples in SPSS, R, SAS, and Stata.
-- PDF handouts of presentation slides. Made available ahead of each session so you can download, print out, and take notes as you follow along.
-- Video screen capture recordings of each workshop session. Made available within 48 hours after each session so you can review the material at your convenience. If you miss any of the live sessions, you can still participate on your own schedule.
-- Exercises. (yes, HOMEWORK!) You really need to practice this stuff and get your hands dirty, so we’re giving you the data to try it on your own. But don’t worry–you won’t be on your own stymied by some coding error that won’t work. You’ll get the code to do the exercises and the answers in case you get stuck.
-- A page to submit written questions between sessions. Got a question as you’re reviewing the video recording or your notes? Just submit a question in the workshop center. Jeff will answer it there if he can, or if it’s something he needs to show you, he’ll answer it in the next Q&A session.
-- Video recordings of all Q&A sessions to review at your convenience. You can even submit questions for the Q&A sessions ahead of time, and Jeff will answer them in the next session and you can watch at your convenience.
-- A list of helpful resources and suggestions for further reading. There’s no required textbook, but there are some good resources that we recommend to support your learning.
-- Bonus videos. We’ve included a few videos from some webinars on relevant topics to help your understanding. Included are:
> A Review of Logarithms for the Data Analyst
> Logistic Regression for Count and Proportion Data
> Zero Inflated Models
> Working with Truncated and Censored Data
You’ll have access to this site and all the related materials and resources for ONE FULL YEAR. That means you can re-watch sessions and ask additional questions again during that 12-month period.
Module 1: Understanding Count Models
Then we’ll go through a brief overview of the most important concepts and steps so you have a big-picture understanding. This lays a strong foundation for the rest of the workshop.
-- Discrete vs Continuous Variables
-- The Variety of Count Data
-- Why OLS Linear Regression Doesn’t Work with Count Data
-- Modeling Assumptions for Count Data
-- The Modeling Process
-- Offset/Exposure Variables
-- Incidence Rate Ratios
Module 2: The Poisson Model
Now that you’ve had an overview, we’ll dig deep into the simplest count model: the Poisson. We’ll explore the issues, terminology, and modeling that apply to all count models within the context of this simplest of models.
-- Important Terminology
-- Poisson Model Assumptions
-- Poisson Distribution
-- Analyzing Data and Model Fit
-- Interpreting the Results
-- Marginal Effects
Now that you understand the Poisson model, we expand on it. Unfortunately, the Poisson model often doesn’t fit real data. In this module we introduce the powerhouse of count models: the negative binomial.
-- Negative Binomial Distribution
-- Important Terminology
-- When to use Negative Binomial Models
-- Running Models
-- Model Fit
-- Negative Binomial Model Assumptions
Module 4: Model Diagnostics and Truncated Models
-- Predicted Values and Residuals: comparison of models
-- Influential Observations: Cook’s distance
-- Residuals versus Predicted Values
-- Residuals versus Predictors
-- Zero Truncated Models
Module 5: Hurdle and Zero-inflated Models
-- Hurdle Model
-- Zero Inflated Poisson Model
-- Zero Inflated Negative Binomial Model
Module 6: Extension and Review
First, we’ll introduce some lesser-known, but valuable alternative models that are sometimes exactly what you need.
There are 6 Workshop Webinar Sessions.
The schedule is as follows:
Workshop orientation (Optional) - Monday, July 15, at 2pm EDT
Module 1 – Tuesday, July 16 at 2pm EDT
Module 2 – Thursday, July 18 at 2pm EDT
Module 3 – Tuesday, July 23 at 2pm EDT
Module 4 – Thursday, July 25 at 2pm EDT
Module 5 – Tuesday, July 30 at 2pm EDT
Module 6 – Thursday, August 1 at 2pm EDT
We will also meet for four, 60-minute Q&A sessions throughout this time. We will meet on the following dates:
Q&A 1 – Wednesday, July 17 at 2pm EDT
Q&A 2 – Wednesday, July 24 at 2pm EDT
Q&A 3 – Wednesday, July 31 at 2pm EDT
Q&A 4 – Wednesday, August 14 at 2pm EDT
Remember, everything is recorded and available for you to watch at your convenience, should you be unable to attend a live session.
Also, you have access to all the class materials for a full 12 months from your enrollment date.
Every Live Workshop from The Analysis Factor includes:
-- All the data, programming code, exercises and handouts you’ll need
-- Live lessons and question & answer sessions
-- Video recordings of all live content for later or repeated review
--The opportunity to ask questions live and to submit questions to be answered between sessions
I have taught this workshop in the past and have watched students grow in confidence and master count models.
My interests are wide-ranging. I work with linear, binomial, count, and mixed models. In addition my work can entail running exploratory and confirmatory models, Structural Equation Models (SEM), latent class analysis, and multiple imputation models for missing data.
I understand that to be an effective instructor, it takes more than subject matter knowledge and a logical approach to analyzing data. I truly enjoy working with people, and I care about your success!
So what kind of background in statistics do you need?
We’re assuming you understand:
-- Least-squares estimation
-- Dummy coding
Familiarity with logistic regression will be helpful, but it’s not necessary.
If you have questions about whether you’re ready for this class, just email us. We’ll give you our honest opinion. We want you to succeed!
“Loved the real world examples. [The course included] not only useful but practical information.”
“The guidance of when and when not to use the different count models available can make a big difference.”
-- Martin Watts, FORCOMP Forestry Consulting Ltd
“The workshop covers a range of useful situations for modeling count data. Thus, it provided the opportunity to work with real data, and real count data issues. At the end of the workshop, one feels as if their analytics tool box can handle a wider variety of jobs.”
-- Carl H. - Graduate Student
Your Satisfaction Is Guaranteed
As with all our programs, your satisfaction is guaranteed. If you participate fully in this workshop–watch, read, and try out everything included–and find you are not satisfied for any reason, we will give you a full refund, no questions asked. Just notify us within 90 days of purchasing the program.