PSYED 3408Syllabus

PSYED3408: Hierarchical Linear Modeling

School of Education

University of Pittsburgh

Spring, 2016

Instructor:Feifei YeLocation:5520B WWPH

Time:12:00-2:40 WednesdayOffice: 5924 WWPH

Phone:412-624-7233Email:

Office Hours: 11-12pm, Tuesday

Course Overview:

The purpose of this course is to introduce hierarchical models for continuous and discrete outcome. Hierarchical models are used when the units of observation are grouped within clusters. In such clustered data observations for the same cluster cannot be assumed to be mutually independent for given covariate values as required by conventional linear and logistic regression. Longitudinal or repeated measures data can also be thought of as clustered data with measurement occasions clustered within subjects. This course will focus on understanding the hierarchical (generalized) linear models and their assumptions, as well as practical aspects of developing models to address research questions and interpreting the findings. This course emphasizes practical, hands-on development, analysis and interpretation of hierarchical linear and nonlinear models. Applications will be drawn from education, psychology, and other social sciences disciplines.

Prerequisites

PSYED 2410 (Applied Regression) or equivalent

Text

Required:

Raudenbush, S. W. Bryk, A. S. (2002). Hierarchical linear models: Applications and data analysis methods, 2nd edition. Newbury Park, CA: Sage.

Lecture notes and handouts: Lecture notes (copies of slides presented in class), handouts, and additional articles or monographs on relevant topics will be made available on the course web.

Recommended:

Raudenbush, Bryk, Cheong, & Congdon (2010). HLM7 (manual).Lincolnwood, IL: SSI.

Supplementary texts

Singer & Willett (2003). Applied Longitudinal Data Analysis. NY: Oxford

Snijders, T., & Bosker, R. (2011). Multilevel analysis: An introduction to basic and advanced multilevel modeling. London: Sage.

More resources on multilevel modeling at

Computing

We will mainly use the computer program—HLM—for this course. Copies of the software are available on the computers in the computer lab in 5520 WWPH. Additional information about the software and HLM in general can be found at the website for Scientific Software International ( The company also provides a free student-version of the software ( The student edition of the software can handle 8,000 level-1 units and 350 level-2 units in a two-level model, or 8,000 level-1 units, 1,700 level-2 units, and 60 level-3 units in a three-level model. No more than 5 predictors may be included at any level of the model, and no more than 25 effects may be included in the whole model. Student version will be sufficient for assignments.

Students taking this course should be comfortable preparing and analyzing data in SPSS. Through this course, we will use SPSS as well for data preparation and some data analysis. It is fine if you choose other software for assignments, but questions regarding your software will not be discussed in class.

Evaluation

Homework: There are a total of five assignments. All assignments turned in after the due date will be docked 20% of the assignment total for each day late. Extensions will be granted only in the case of personal emergency.

Article Critique:You will review one article published in a journal in your field that adopted hierarchical linear model following the reviewer guideline provided on the course web. You will present your critique of the article at the beginning of one class. You will sign up for the date you are going to present at the beginning of the term.

Term Project: The purpose of the term project is to provide experience in using, interpreting and reporting the results of HLM models. The paper must represent the original analysis of data that you have not done before. This does not mean that you cannot use existing data or use a study on which you have previously conducted analyses; it means you need to conduct new analyses not attempted before. For the data analysis project, you need to

  • Identify a research problem that can be studied through hierarchical analysis.
  • Select a hierarchical data set that can be used to study the problem. You can use either your own data or a public dataset.
  • Construct appropriate variables to use in the analysis.
  • Develop appropriate HLM models to test the research problem.
  • Test the models using the HLM program.
  • Describe and interpret the results.

On 3/23/2016, you will turn in an electronic copy of proposal for approval which addresses the first three questions above. The proposal should be 3single-space pages or less (reference not included).Based on the feedback you receive, you will revise this proposal and perform the analyses. The final report, due on 4/20/2016, will provide a brief literature review, research questions, methods, the results of your analysis, your conclusion and discussion. The final report should be no longer than 15 double-spaced pages, tables and figures excluded. The output as well as the commands used to prepare and analyze the data should also be provided.

You will present your term project on 04/20/2016 or 04/27/2016. The presentation, which should be around 25 minutes in length, should (at least) include introduction (research context and questions), methods (sample, variables, and statistical models), results, and discussion.

You will be evaluated on the basis of your performance on the homework assignments (20% of your grade), the midterm exam (30% of your grade), the article critique (10% of your grade), and the term project (40% of your grade, 30% for the report and 10% for the presentation).

Letter grades will be based on actual points earned as follows:

Point / Letter / Point / Letter
Above 96 / A+ / 77-80 / C+
93-96 / A / 74-77 / C
90-93 / A- / 70-74 / C-
87-90 / B+ / 67-70 / D+
84-87 / B / 64-67 / D
80-84 / B- / 60-64 / D-
Below 60 / F

For students who choose to audit, there is no requirement and it is up to you regarding whether to complete assignments, midterm exam, article critique, or term project.

Tentative Course Outline

Week / Date / Topic / Reading (R&B) / HW
1 / 01/06/2016 / Course overview / Chapter 1
2 / 01/13/2016 / Random intercept models / Chapters 2 (p16-31) & 4 (p68-75) / HW1
3 / 01/20/2016 / Random slope models and beyond / Chapters 4 (p75-85) / HW1 due
4 / 01/27/2016 / Estimation
Centering / Chapter 3 (p85-94)
Chapters 2 (p31-35) & 5 (p134-149) / HW2
5 / 02/03/2016 / Model building / Chapter 4 (p149-152)
Chapters 9 (p252-263, p267-273) / HW2 due
6 / 02/10/2016 / Assumptions and diagnostics / Chapters 9 (p263-267, p273-280) / HW3
7 / 02/17/2016 / Two-level growth models (I) / Chapter 6 (p160-185) / HW3 due
8 / 02/24/2016 / Two-level growth models (II) / Chapter 6 (185-199)
9 / 03/02/2016 / In-class Midterm-Exam
10 / 03/09/2016 / No class (spring break)
11 / 03/16/2016 / Three-level models
Power analysis / Chapter 8 / HW4
12 / 03/23/2016 / Hierarchical generalized linear model (I) / Chapter 10 (p291-309) / HW4 due
Project Proposal due
13 / 03/30/2016 / Hierarchical generalized linear model (II) / Chapter 10 (p309-335) / HW5
14 / 04/06/2016 / Cross-classified model / Chapter 12 / HW5 due
15 / 04/13/2016 / Missing data
16 / 04/20/2016 / In-Class presentation of final project / Term Project Due
17 / 04/27/2016 / In-Class presentation of final project

1