An Introduction to HLM and SEM

Slides:



Advertisements
Similar presentations
Questions From Yesterday
Advertisements

Using Growth Models to improve quality of school accountability systems October 22, 2010.
Hierarchical Linear Modeling: An Introduction & Applications in Organizational Research Michael C. Rodriguez.
Qualitative predictor variables
CORRELATION. Overview of Correlation u What is a Correlation? u Correlation Coefficients u Coefficient of Determination u Test for Significance u Correlation.
Statistical Analysis Overview I Session 2 Peg Burchinal Frank Porter Graham Child Development Institute, University of North Carolina-Chapel Hill.
Outline 1) Objectives 2) Model representation 3) Assumptions 4) Data type requirement 5) Steps for solving problem 6) A hypothetical example Path Analysis.
Multiple Linear Regression Model
Multiple Regression Models. The Multiple Regression Model The relationship between one dependent & two or more independent variables is a linear function.
When Measurement Models and Factor Models Conflict: Maximizing Internal Consistency James M. Graham, Ph.D. Western Washington University ABSTRACT: The.
Agenda for January 25 th Administrative Items/Announcements Attendance Handouts: course enrollment, RPP instructions Course packs available for sale in.
Basic Business Statistics, 11e © 2009 Prentice-Hall, Inc. Chap 14-1 Chapter 14 Introduction to Multiple Regression Basic Business Statistics 11 th Edition.
Topic 3: Regression.
1 1 Slide © 2015 Cengage Learning. All Rights Reserved. May not be scanned, copied or duplicated, or posted to a publicly accessible website, in whole.
Correlational Designs
Correlation 1. Correlation - degree to which variables are associated or covary. (Changes in the value of one tends to be associated with changes in the.
Chapter 7 Correlational Research Gay, Mills, and Airasian
Quantitative Research
Latent variables in psychology and social sciences: theoretical positions, assumptions, and methodological conundrums Alina Zlati.
SPSS Statistical Package for Social Sciences Multiple Regression Department of Psychology California State University Northridge
Analysis of Clustered and Longitudinal Data
Statistics for Managers Using Microsoft Excel, 4e © 2004 Prentice-Hall, Inc. Chap 13-1 Chapter 13 Introduction to Multiple Regression Statistics for Managers.
Structural Equation Modeling in Human-centric Computing Research: A Study of Electronic Communication in Virtual Teams Using WarpPLS Ned Kock Texas A&M.
3. Multiple Regression Analysis: Estimation -Although bivariate linear regressions are sometimes useful, they are often unrealistic -SLR.4, that all factors.
Chapter 12 Correlation and Regression Part III: Additional Hypothesis Tests Renee R. Ha, Ph.D. James C. Ha, Ph.D Integrative Statistics for the Social.
Multivariate Statistics Principal Components/Factor Analysis Structural Equation Modeling.
1 FORECASTING Regression Analysis Aslı Sencer Graduate Program in Business Information Systems.
From GLM to HLM Working with Continuous Outcomes EPSY 5245 Michael C. Rodriguez.
1 1 Slide © 2003 Thomson/South-Western Chapter 13 Multiple Regression n Multiple Regression Model n Least Squares Method n Multiple Coefficient of Determination.
1 1 Slide © 2007 Thomson South-Western. All Rights Reserved Chapter 13 Multiple Regression n Multiple Regression Model n Least Squares Method n Multiple.
1 1 Slide Multiple Regression n Multiple Regression Model n Least Squares Method n Multiple Coefficient of Determination n Model Assumptions n Testing.
Hierarchical Linear Modeling (HLM): A Conceptual Introduction Jessaca Spybrook Educational Leadership, Research, and Technology.
Chapter 14 Introduction to Multiple Regression
Introduction Multilevel Analysis
L 1 Chapter 12 Correlational Designs EDUC 640 Dr. William M. Bauer.
Educational Research: Competencies for Analysis and Application, 9 th edition. Gay, Mills, & Airasian © 2009 Pearson Education, Inc. All rights reserved.
Multiple Linear Regression: Cloud Seeding By: Laila Rozie Rozie Vimal Vimal.
Multiple Linear Regression. Purpose To analyze the relationship between a single dependent variable and several independent variables.
Introduction to Multilevel Modeling Stephen R. Porter Associate Professor Dept. of Educational Leadership and Policy Studies Iowa State University Lagomarcino.
Measurement Models: Exploratory and Confirmatory Factor Analysis James G. Anderson, Ph.D. Purdue University.
Chapter 13 Multiple Regression
Aron, Aron, & Coups, Statistics for the Behavioral and Social Sciences: A Brief Course (3e), © 2005 Prentice Hall Chapter 12 Making Sense of Advanced Statistical.
EDCI 696 Dr. D. Brown Presented by: Kim Bassa. Targeted Topics Analysis of dependent variables and different types of data Selecting the appropriate statistic.
Impediments to the estimation of teacher value added Steven Rivkin Jun Ishii April 2008.
Data Analysis in Practice- Based Research Stephen Zyzanski, PhD Department of Family Medicine Case Western Reserve University School of Medicine October.
Regression Analysis: Part 2 Inference Dummies / Interactions Multicollinearity / Heteroscedasticity Residual Analysis / Outliers.
Chapter 15 The Chi-Square Statistic: Tests for Goodness of Fit and Independence PowerPoint Lecture Slides Essentials of Statistics for the Behavioral.
Basic Business Statistics, 10e © 2006 Prentice-Hall, Inc.. Chap 14-1 Chapter 14 Introduction to Multiple Regression Basic Business Statistics 10 th Edition.
American Educational Research Association Annual Conference New York – March 24-28, 2008 Noelle Griffin, Ph.D. Evaluation of an Arts-Based Instructional.
Problem What if we want to model: –Changes in a group over time Group by time slices (month, year) –Herd behavior of multiple species in the same area.
Statistics for Managers Using Microsoft Excel, 5e © 2008 Prentice-Hall, Inc.Chap 14-1 Statistics for Managers Using Microsoft® Excel 5th Edition Chapter.
Applied Quantitative Analysis and Practices LECTURE#28 By Dr. Osman Sadiq Paracha.
Chapter 17 STRUCTURAL EQUATION MODELING. Structural Equation Modeling (SEM)  Relatively new statistical technique used to test theoretical or causal.
Methods of multivariate analysis Ing. Jozef Palkovič, PhD.
Methods of Presenting and Interpreting Information Class 9.
EXPERIMENTAL RESEARCH
Structural Equation Modeling using MPlus
Consequential Validity
Moderation, Mediation, and Other Issues in Regression
An introduction to basic multilevel modeling
Shudong Wang NWEA Liru Zhang Delaware Department of Education
Lecture 4 - Model Selection
12 Inferential Analysis.
Making Sense of Advanced Statistical Procedures in Research Articles
From GLM to HLM Working with Continuous Outcomes
Structural Equation Modeling
12 Inferential Analysis.
Day 49 Causation and Correlation
HW# : Complete the last slide
Multiple Linear Regression
Presentation transcript:

An Introduction to HLM and SEM Carolyn Furlow

Hierarchical Linear Modeling (HLM) Structural Equation Modeling (SEM) Multilevel models or Hierarchical Linear Models and Structural Equation Models are both considered extensions of regression analyses. Both are frequently used with educational data and are rapidly gaining in popularity.

When should HLM be used? HLM is appropriate for use when we have nested data structures which occurs frequently with educational data. For example, when we have students who are nested in classrooms, classrooms nested within schools, etc… E.g., if we randomly sampled three classrooms of students from 10 different schools and then collected data from all these students.

Other HLM scenarios with nested data Clients in groups for group therapy Employees in organizations School administrators in school districts Voters in voting precincts Homeowners in neighborhoods

Unit of Analysis Researchers have difficulty deciding the appropriate unit of analysis with educational data. Should the student be the unit of analysis or the classroom mean, school mean, etc.? HLM simultaneously accounts for several levels of data E.g., if we have a multiple regression model do we try to predict the student’s level of achievement or the classroom’s?

HLM uses We can simultaneously study the effects of group level variables and individual level variables with HLM There may be interactions across levels as well that only HLM can account for. For example, the effect of student study time may be related to teacher emphasis on homework.

Why not just use multiple regression? Students from Classroom A tend to be more alike with each other than they would be with students from Classroom B. Students within any one classroom, b/c they were taught together tend to be similar in their performance As a result, they provide less information than if the same number of students had been taught separately by different teachers Problem is if student is used as the unit of analysis then we have problems with error. If classroom used then we have a loss of power

Why not just use multiple regression? Therefore the assumptions of constant variance and independence of errors in multiple regression are violated. Incorrect standard errors and tests of significance for regression coefficients would be given using MR when HLM should be used.

Example from Tate Example of a policy analysis related to ongoing school reform efforts in a hypothetical state. Set of instructional objectives for fifth grade science were developed but individual schools not required to use objectives in their curriculum

Example from Tate Annual state-wide test was modified to reflect the new objectives Evidence that individual schools vary with respect to how consistent their science classes are with objectives

Example from Tate Policy makers have several research questions Question 1 (group level) Is the average school achievement on the state-wide science test, controlling for student aptitude, related to the degree to which the school science instruction is consistent with the state-wide objectives?

Example from Tate Question 2 (individual level) Is the relationship between individual science achievement and individual aptitude within each school related to the degree to which the school science curriculum is consistent with the state-wide objectives?

Hypothetical Study Random sample of 20 schools from the state Collected measures of individual science achievement and aptitude for all 580 students in the 20 schools Each school has also been given a score on a scale reflecting “Degree of Consistency of School Science Instruction with State-Wide Objectives”

Hypothetical Study We can test at the group level how much the school’s level of consistency affects the variability of school’s scores on the achievement test We can also test whether the relationship between individual achievement and aptitude is related to how consistent the curriculum is with the objectives

Structural Equation Modeling (SEM) Also seen as an extension of regression analysis. SEM attempts to analyze more complicated causal models and can incorporate unobserved (latent) variables and mediating variables as well as observed (measured) variables SEM involves imposing a theoretical model on a set of variables to explain their relationships.

SEM Latent variables are unobserved/unobservable variables such as self-esteem, marital happiness, depression. These are sometimes called factors. They are measured by indicators (observed variables), often behaviors that can be observed such as stated chance of getting divorced, number of fights with spouse in the last week.

SEM Standard SEM – consists of mediating variables and latent variables Special Cases of SEM Path analysis - all variables are observed but some type of mediating variable exists Confirmatory factor analysis - where a latent variable such as intelligence is measured by several indicator variables

SEM Obtain overall test of how well our data fits with our proposed model Also obtain significance values for each of the paths between variables

Example of SEM (path-analytic model) Authoritative Parenting Style Ethnic Identity Family Stress Global Self-esteem All observed variables in this model. This is a special case of SEM. SEM can also accommodate latent variables and confirmatory factor analysis. Basically entails specifying entire model before running it. Then model fit can be assessed and whether you need to go back and respecify the model. Thoroughly explain the model. Interrelatedness and medational variable. Overidentified? Ethnic identity-APS plays a role in GSE b/c it works through ethnic identity Teacher Support

Confirmatory Factor Analysis

SEM