Online course: Latent Class (LC) Discrete Choice Modeling with Scale Factors
Topics include:
- Introduction to LC choice modeling
- First Choice and Best-Worst models
- LC choice modeling with scale factors
- Developing MaxDiff typing tools
COURSE CONTENT
Discrete choice models are widely used to forecast market share and to design optimal products and services. Latent Class (LC) models are natural tools to analyze traditional first choice, Best-Worst, and other choice data to identify segments having different preferences. However, due to the difficulty in isolating scale effects, current segmentation models often obtain segments that may differ primarily in scale (response error) rather than real preference patterns. This course will introduce the basics of LC Choice modeling and Best-Worst (MaxDiff) scaling models and describe the latest approaches for dealing with scale confounds — Scale Adjusted Latent Class (SALC) models and Latent Class Tree (LCT) models.
SALC modeling allows for the adjustment for the effects of scale to assure that the resulting segments are more meaningful, differing only in preference, while the LCT modeling helps with segment interpretation by allowing for and structuring substantively meaningful “root” or “theme” classes.
COURSE STRUCTURE
The course will cover how to set up choice models in Latent GOLD using the Basic (GUI) and Syntax modules. Participants will have access to the demo version of Latent GOLD 5.1 Choice and will not need a current license to participate in the course.
The course is taught by Dr. Jay Magidson, co-developer of Latent GOLD® and founder of Statistical Innovations. This online course includes lecture notes, readings, and hands-on exercises using Latent GOLD. Course participants will be given access to a private bulletin board that serves as a forum for discussing of ideas, solving problems, and interacting with the instructor. The courses take place over three weeks, and should require about 5-10 hours per week. Material for each week is posted on the Friday before each week, along with answers to the previous week’s exercises.
For only $495 ($295 for university professors and students), you will receive access to the private course bulletin board, course website, unique problem sets and assigned reading. Tuition is half-price for additional participants from the same organization. Attendees who
participated in last year’s choice modeling with scale factors course may join us at a 40% discount.
Description
Course Overview: Discrete choice models are widely used to forecast market share, design optimal products and services and more. Latent Class (LC) models are natural tools to analyze traditional first choice only, MaxDiff and other rankings data to identify segments with differing preferences. However, due to the difficulty in isolating scale effects, current segmentation models often obtain segments that differ primarily in scale (response error) rather than real preference patterns. This course will introduce the basics of LC Choice modeling, MaxDiff (best-worst scaling) models, and Anchored MaxDiff models. The course will also describe the latest approaches for dealing with scale confounds — Scale Adjusted Latent Class (SALC) models and Latent Class Tree (LCT) models.
SALC modeling allows for the adjustment for the effects of scale to assure that the resulting segments are more meaningful, differing only in preference, while the LCT modeling helps with segment interpretation by allowing for and structuring substantively meaningful “root” or “theme” classes.
Course Structure: The course takes place online. Course participants will be given a username and password for access to a private bulletin board that serves as a forum for discussion and interaction with the instructor. The course is divided into three weekly sessions. Attendees typically spend about 5-15 hours on each session. At the beginning of each week, participants receive the relevant material, in addition to answers to exercises from the previous session. All course materials are posted to a dedicated course homepage, which can be accessed via the same username and password. During each session, participants review the course materials and work through exercises using the Latent GOLD® Choice program. The instructor will provide answers to the exercises and to posted questions, but participants may also engage in discussions with other course participants.
Course program:
Session 1: Introduction to Choice Modeling with Scale Factors
- Scale-Adjusted Latent Class (SALC)
- Introductory Topics
- Choice modeling basics (MNL with first choice data)
- Latent classes (Choice tutorials 1 & 2)
- Introduction to scale factors
- The general log-scale model
Session 2: Putting it All Together — SALC Models in Practice
- Modeling Scale Factors using observed vs. latent variables (scale classes)
- First choice models
- MaxDiff models
- Allowing different scale factors for best and worst
- Including other observed covariates (time to complete)
- Models with both observed and latent scale heterogeneity
Session 3: Advanced SALC Topics
- Modeling continuous scale heterogeneity (latent continuous scale factor)
- Practical guidelines in estimating SALC models
Who should sign up for this course: This course is intended for researchers and practitioners in marketing, economics, and fields in which consumer demand and choice is of interest.
Prerequisite: Familiar with the use of the standard multinomial logit model with first choice data. This prerequisite can be met by reviewing LG Choice Tutorial #1 (or LG Choice Tutorial #2 ) with the demo version of Latent GOLD®.
Course Material: No text required — copies of published articles and other material will be made available. All participants will have free access to the demo version of Latent GOLD® Choice, which allows unrestricted analyses of all course datasets. Participants need not license a copy of the Latent GOLD® Choice program. All participants will have free access to the demo version of Latent GOLD® Choice, which allows unrestricted analyses of all course datasets.
Instructor: Dr. Jay Magidson, founder and president of Statistical Innovations Inc.. Dr. Magidson’s clients have included A.C. Nielsen Co., Household Finance Corp., Blue Cross Blue Shield Association, and Pfizer. He taught statistics at Tufts and Boston University and is widely published on the theory and applications of multivariate statistical methods. Dr. Magidson designed SPSS CHAID, SI-CHAID®, GOLDMineR® and CORExpress®, is the co-developer (with Jeroen K. Vermunt) of the Latent GOLD® and Latent GOLD® Choice programs.