Short Courses in Detroit, Michigan, June 2016 - Two Sessions, Fourteen Courses
Week 1: Quantitative Research Methods
June 6-11, 2016 - Two Sessions, Fourteen Courses
Session 1: June 6-8, Seven Courses | Session 2: June 9-11, Seven Courses
Short Course Sessions and Groupings
We offer two sessions which allows course participants the opportunity to take two back-to-back courses that compliment one another. All courses in a session are taught concurrently, so a participant can take only one course per session.
Short Course Topics, Instructors and Summaries
Complete Course Listing
Click a course name to see more info.
Session 1: June 6-8, Seven Courses
“Introduction to Structural Equation Methods"
The Introduction to Structural Equation Methods Short Course provides (a) introductory coverage of latent variable techniques, including confirmatory factor analysis and structural equation methods with latent variables, (b) discussion of special issues related to the application of these techniques in organizational research, and (c) a comparison of these techniques with traditional analytical approaches. This Short Course will contain a balance of lecture and hands-on data analysis with examples and assignments, and emphasis will be placed on the application of SEM techniques to organizational research problems. Participants will:
- develop skills required to conduct confirmatory latent variable data analysis, based on currently accepted practices, involving topics and research issues common to organizational research.
- learn the conceptual and statistical assumptions underlying confirmatory latent variable analysis.
learn how to implement data analysis techniques using software programs for confirmatory modeling. Special emphasis will also be placed on the generation and interpretation of results using the contemporary software programs LISREL, MPlus, and Amos.
- learn how latent variable techniques can be applied to contemporary research issues in organizational research.
- learn how the application of current latent variable techniques in organizational research differs from traditional techniques used in this literature
- complete in-class exercises using their preferred package (LISREL, MPlus, and Amos)
Required Software: LISREL (free trial edition), MPlus or Amos
Note: For those without current access to an SEM package, LISREL has a free trial edition that you should download no earlier than 1 week before class. The MPlus demo is not adequate for this course.
"Advanced SEM I: Measurement Invariance, Latent Growth Modeling & Nonrecursive Modeling"
The short course covers three advanced structural equation modeling (SEM) topics: (a) testing measurement invariance; (b) latent growth modeling; and (c) evaluating reciprocal relationships in SEM. The instructor uses the Mplus SEM software package throughout the workshop. To get maximum benefit from this short course, the participants should have the full version of Mplus loaded on their laptops
and bring the laptop with them to the course. The instructor lectures about half of the time with the remaining time devoted to having participants run examples with actual data provided by the instructor. Participants go home with usable examples and syntax. The measurement invariance testing section focuses on the procedures as outlined in the Vandenberg and Lance (2000) Organizational Research Methods article. Namely, we will cover the 9 invariance tests starting with the tests of equal variance-covariance matrices and ending with tests of latent mean differences. Other outcomes of covering these tests are how to use Mplus syntax, how to do multi-sample analyses, and also how to test hypothesized (a priori) group mean differences but using the latent means of the latent variables within each group. Thus, the first section accomplishes much more than the just the measurement invariance tests. The workshop then advances to operationalizing latent growth models within the SEM framework. Essentially, this is how to use one's longitudinal data to actually capture the dynamic processes in one's theory. Thus, it is very, very different than cross-sectional tests where one is stuck in one point in time. Again, this is what goes on at the surface level, but the participant will also be exposed to modeling how the change in one variable impacts change in another. We will also use mixed modeling. And at the end of it, I introduce the participants to latent profile modeling with latent growth curves. The final piece is the testing of models with feedback loops via an SEM-Journal article by Edward Rigdon (1995). We will go through his 4 different models and what they mean.
Required Software: MPlus (order the full version, try the free demo version)
"Mediation and Moderation"
Mediators and moderators are pervasive in organizational research. As theoretical frameworks advance, the specification and testing of mechanisms (or mediators) through which the effects of one variable on another are transmitted has become a central concern. Another primary concern of theory development is the specification and testing of boundary conditions (or moderators) that affect the nature (significance, magnitude and/or direction) of the relationship between variables. More recently there has been increased attention on the theoretical specification of models that incorporate both mediation and moderation, facilitated in no small part by advancements made in the analytical approaches for testing such effects. This course will provide an overview of the conceptual, methodological and statistical issues surrounding the investigation of mediators and moderators, as well as their combined dynamics. The short course will feature both lecture and hands-on components, including analytical demonstrations and exercises that focus on the testing (primarily using SPSS) of these effects.
Required Software: SPSS (free trial version)
"Introduction to Multilevel Analysis"
This course is aimed at faculty and students who are relatively new to multilevel theory, measurement, and analysis. It will review basic issues associated with the development and testing of multilevel theories. Although the focus will be on issues pertaining to multilevel theory and measurement (e.g., multilevel constructs, multilevel construct validation, aggregation and composition models), we will also discuss general issues associated multilevel analysis. Examples will be presented and discussed using both SPSS and HLM. The R package will be introduced, explained, and emphasized during this short course in preparation for the advanced short course in Session II. Specific topics will include:
- Module 1: Multilevel Theory: Constructs, Inferences, and Composition Models
- Module 2: Multilevel Measurement: Aggregation, Aggregation Bias, & Cross-Level Inference
- Module 3: Multilevel Measurement: Estimating Interrater Agreement & Reliability
- Examples using SPSS Software
- Module 4: Multilevel Measurement and Multilevel Modeling: A Simple Illustration of Analyzing Composite Variables in Hierarchical Linear Models
- Examples using SPSS, HLM, and R Software
- Module 5: Wrap up and Final Q & A
Required Software: R (
), SPSS (
free trial version
), HLM (
"Introduction to R"
This course will provide a gentle introduction to the R computing platform and the R-Studio interface. We will cover the basics of R such as importing and exporting data, understanding R data structures, and R packages. You will also learn strategies for data manipulation within R (compute, recode, selecting cases, etc.) and best practices for data management. We will work through examples of how to conduct basic statistical analyses in R (descriptive, correlation, regression, T-test, ANOVA) and graph those results. Finally, we will explore user-defined functions in R and lay the groundwork for understanding how to perform more complex analyses presented in later CARMA short courses.
Required Software: R (download here)
“Meta-Analysis and Systematic Reviews”
This course provides the participant with knowledge concerning the major meta-analysis models used in research in organizational science and other sciences. The course also details all steps in conducting a systematic review. Thus, this course is not solely a statistics/methods course but provides the participant with knowledge needed to conduct a meta-analysis and systematic review consistent with the Meta-Analysis Reporting Standards (MARS). Free software is made available to the participants and hands-on practice in the software is incorporated into the course. The course also addresses emerging topics in meta-analysis and systematic reviews including a detailed treatment of publication bias.
Required Software: R, RStudio *Note: Install R before installing RStudio*
"Introduction to Longitudinal Data Analysis
Much of organizational research is arguably about causal processes of change even as many theories and analyses of change are deficient in the extent to which they specify or test temporal changes. There are many different approaches for measuring change. Each approach embodies a set of assumptions about what change is, what constitutes evidence of change, and some methods are oblivious to evidence of other types of change. This workshop is an introduction to longitudinal analyses.
We will cover how to properly specify a theory regarding change over time and discuss designs for collecting longitudinal data, as well as examine the critical role of measurement and construct validity in longitudinal data. The focus will be on multiple approaches for analyzing longitudinal data. Approaches covered in this workshop include using unconditional change scores and static scores, panel models and repeated measures ANOVA and MANOVA. Newer approaches including, latent growth modeling and random coefficient modeling, will be introduced. The emphasis will be on hands on learning with longitudinal data sets.
Required Software: SPSS (download free trial) and LISREL (order the full version, request free trial*you must contact the developer, SSI, in order to receive your free trial.* *This process can take 1-2 business days.* *The LISREL free student version will not be suitable for this course.*)
Session 2: June 9-11, Seven Courses
“Intermediate SEM: Model Evaluation"
Dr. Larry J Williams, Wayne State University
This course is aimed at faculty and students with an introductory understanding of structural equation methods who seek a better understanding of the challenging process of making judgments about the adequacy of their models. Those who attend should have experience in fitting structural equation models with software such as LISREL, MPlus, EQS, or AMOS. This experience requirement can be met by completion of the Introduction to SEM Short Course. Attendees will be expected to bring their own laptop computers installed with their SEM software, and they should also know how to import data from an SPSS save file into their SEM software program. Attendees will learn out to interpret and report results from SEM analyses, and how to conduct model comparisons to obtain information relevant to inferences about their models, as well as advantages and disadvantages of different approaches to model evaluation. Attendees are encouraged to bring their own data for use during parts of the short course.
The course will consist of five sections, with each section having a lecture and lab component using exercises and data provided by the instructor:
• Review of model specification and parameter estimation
• Overview of model evaluation
• Logic and computations for goodness-of-fit measures
• Analysis of residuals and latent variables
• Model comparison strategies
Required Software: Your preferred SEM software package, SPSS (free trial version)
"Advanced SEM II: Missing Data Issue in SEM, Multi-Level SEM and Latent Interactions"
The workshop covers three advanced structural equation modeling (SEM) topics: (a) multilevel modeling; (b) latent interactions; and (c) dealing with missing data in SEM applications. The instructor uses the Mplus SEM software package throughout the workshop. To get maximum benefit from this short course, the participants should have the full version of Mplus loaded on their laptops
and bring the laptop with them to the course. The instructor lectures about half of the time with the remaining time devoted to having participants run examples with actual data provided by the instructor. Participants go home with usable examples and syntax. The multilevel modeling section starts out using observed variables only, and no latent variables. Parallels are drawn in this approach and the other packages such as HLM. The main purpose here, though, is to teach participants the basics of multilevel modeling such as aggregation, cross-level interactions and cross-level direct effects. The workshop advances to using latent variables in a multi-level environment. Particular focus will be on multilevel confirmatory factor analysis whereby separate measurement models are estimated at both the within and between levels. The topic then switches to multilevel path modeling with emphasis on between vs. within modeling, and the estimation of cross-level interaction and direct effects among latent variables. The latent interaction section focuses on specifying interactions among latent variables in SEM models. This section starts out with a review of basic interaction testing within a regression environment (using Mplus). From this foundation, participants will move into specifying interactions among latent variables and how to test hypotheses with interactions. And from this point, the workshop will move into moderated-mediation but from the SEM perspective. The final segment of the short course deals with missing data. A great deal of time at the beginning is spent on missing data patterns and why they occur. The workshop then moves into the old methods of dealing with missing data such as listwise and pairwise deletion, and mean or regression based imputation. The disadvantages of those methods are discussed. We then move into covering the newer methods for dealing with missing data, multiple imputation, and full information maximum likelihood. Participants will be showed how to utilize the latter methods in Mplus.
Required Software: MPlus (order the full version, try the free demo version)
“Alternatives to Difference Scores: Polynomial Regression & Response Surface Methods”
For decades, difference scores have been used in studies of fit, similarity, and agreement in organizational research. Despite their widespread use, difference scores have numerous methodological problems. These problems can be overcome by using polynomial regression and response surface methodology to test hypotheses that motivate the use of difference scores. These methods avoid problems with difference scores, capture the effects difference scores are intended to represent, and can examine relationships that are more complex than those implied by difference scores.
This short course will review problems with difference scores, introduce polynomial regression and response surface methodology, and illustrate the application of these methods using empirical examples. Specific topics to be addressed include: (a) types of difference scores; (b) questions that difference scores are intended to address; (c) problems with difference scores; (d) polynomial regression as an alternative to difference scores; (e) testing constraints imposed by difference scores; (f) analyzing quadratic regression equations using response surface methodology; (g) difference scores as dependent variables; and (h) answers to frequently asked questions.
Required Software: TBA
"Advanced Multilevel Analysis"
The CARMA Advanced Multilevel Analysis short course provides both (1) the theoretical foundation, and (2) the resources and skills necessary to conduct advanced multilevel analyses. Emphasis will be placed on techniques for longitudinal data. The course covers both basic models (e.g., 2-level mixed and growth models), and more advanced topics (e.g., 3-level models, discontinuous growth models, and multilevel moderated-mediation models). Practical exercises, with real-world research data, are conducted in R, with accompanying output from MPlus provided for some examples. Participants who prefer SAS, SPSS, or MPlus and have experience with these programs have the option of completing some assignments with these programs. Participants are encouraged to also bring datasets to the course and apply the principles to their specific areas of research. The course is best suited for faculty and graduate students who have at least some foundational understanding of conducting multilevel analyses.
- Module 1: 2-Level Mixed Models: Cross-Level Main Effects & Interactions
- Introduction to multilevel modeling in R and MPlus
- Exercise 1a: Mixed modeling in R
- Exercise 1a: Mixed modeling in MPlus
- Module 2: Analyzing change and growth
- Exercise 2a: Growth modeling in R
- Exercise 2a: Growth modeling in MPlus
- Module 3: Bayes Estimates and lme4
- Bayes Estimates using lme in R
- Specifying models in lme4
- Module 4: Discontinuous growth models
- Module 5: 3-level models; moderated-mediation models
- Examples using R and MPlus
Required Software: R (download here)
“Multivariate Statistics with R”
This course continues the introduction to R from the first session by covering advanced topics related to multivariate statistics. We will cover topics related to data management for multivariate data and will provide an overview of plotting and visualizing multivariate data in R. Specific learning outcomes include learning how to conduct analyses involving:
- Multiple regression and diagnostics
- Binary, multinomial, and ordinal logistic regression
- Exploratory factor analysis and principal components
- Multivariate regression, canonical correlation, and MANOVA
- Topics in statistical computation (e.g., bootstrapping, Monte Carlo simulation)
- Structural equation modeling with the lavaan package
- Reproducible research for quantitative reports
The session will provide participants with some discussion of necessary background knowledge and practical exercises.
Required Software: R (download here), R Studio (download here), and tex (for Windows:
, for OS X http://www.tug.org/mactex/, for Ubuntu/Debian (Linux): apt-get install texlive or
“Big Data and Data Mining with R”
Big data has been a buzzword for several years both in academia and industry. Although the term is vague and is certainly overused, it does encompass some interesting new ideas and unfamiliar analytical techniques. Notable among these is “data mining,” a family of analytical methods for clustering, classifying, and predicting that go a step beyond the statistical methods used by many social science researchers. In this short course, we will discuss the dimensions of big data and the conceptual steps involved in data mining. Students are welcome to bring their own data sets for experimentation on their own, but this is not required.
We will use the open source statistical processing language, R, for most of the work we do in the course. Extensibility is the hallmark of R; its system of add-on packages provides access to an unequaled range of analytical tools and techniques. You do not have to be an expert in R to take this course, although you will find the course easier if you also take the introduction to R offered by CARMA earlier in the week. Prior to the course, I will ask students to install R on their personal computers and review the first few chapters of my free eTextbook, An Introduction to Data Science. Depending on the interests and preferences of the students, we also use the Rattle or R-Studio graphical user interfaces.
The ideal student will have an interest in using R, knowledge of some basic descriptive and inferential statistics, and some curiosity about exploring alternative, empirically driven strategies for analysis of large data sets. No prior experience with data mining is required and students who participate successfully in this short course can expect to learn enough about data mining to begin experimenting with these tools in research or teaching.
Required Software: R (download here), R Studio (download here)
"Developing Psychological Measures: Conceptual, Psychometric, and Practical Guidelines"
Dr. Fred Oswald, Rice University
Short course attendees will be able to develop measures of psychological constructs such as knowledge, attitudes, personality, or interests. Short course topics include construct validity, content and item generation, measurement models (EFA/CFA/IRT), multi-source measurement, and latent growth models. Each topic will emphasize practically useful guidelines and how to address (and prevent) real-world challenges.
Required Software: TBA