Spring
Quarter Professor
Ross L. Matsueda
2018 227 Savery
Hall
Office Hours: Mon, Tue 2-3pm
(& Thur
7:30-8pm Savery 409)
SOCIOLOGY 529/CS&SS 526
Structural Equation Models for the Social Sciences
COURSE DESCRIPTION:
This course introduces covariance
structure analysis, focusing on Jöreskog and Sörbom's LISREL approach.
It begins with the notion of a causal
structure underlying a set of observable moments (covariances). This
notion is illustrated
briefly with path analysis applied to a multiple equation recursive model of
observable
variables. We will discuss the implications of random
measurement error in linear regression models,
discuss the concept of
unobservable variables, and review some elementary principles of classical test
theory. We then introduce
the LISREL model, describing the model in matrix form, and then briefly present
the estimation issue and
use of maximum likelihood estimation and likelihood ratio testing. We will discuss
the LISREL and PRELIS
software programs, and briefly discuss R package’s SEM and lavaan,
Stata’s SEM
package and Muthén’s Mplus Program. We will then examine the identification issue
and examine specific
classes of models, such as
confirmatory factor models, MIMIC models, regression models with latent
variables, and non-recursive
models. We will also cover estimation when observed data are not multivariate
normal, and observed
variables are ordinal, dichotomous, or censored. Time permitting, we will survey other
models used in recent
research in the social sciences, such as models for panel data, growth curve
models,
models for nested data,
mixture models for latent class and trajectory analysis, and models for data
missing
at random.
COURSE OBJECTIVES:
1. Introduce students to
the fundamentals of structural equation modeling, including specification,
estimation,
and testing.
2. Provide students with
a critical understanding of SEM, including the assumptions required to use
them.
3. Provide students with
the tools to use SEM to address important social scientific problems when the
problems and data are
appropriate to the method.
4. Provide students the
opportunity for specifying, estimating, and testing a simple SEM and writing up
the results
in a short paper structured on the model
of a research article.
5. Provide students with
a solid foundation in SEM with which to pick up more advanced topics not
covered
here.
REQUIRED TEXT:
Bollen, Kenneth A. 1989. Introduction to Structural Equation Models
with Latent Variables. New York:
Wiley.
RECOMMENDED:
Hayduk, Leslie. 1987. Structural Equation Modeling with
LISREL: Essentials and Advances. Baltimore:
Johns Hopkins Press.
Jöreskog, Karl G., Ulf H.
Olsson, and Fan Y. Wallentin. 2016. Multivariate Analysis with LISREL.
Switzerland: Springer International
Publishing.
Byrne, Barbara M. 1998. Structural Equation Modeling with LISREL,
PRELIS, and SIMPLIS: Basic Concepts,
Applications, and Programming. Mahwah, NJ: Lawrence Erlbaum.
Syllabus Sociology
529/CSSS 526 Course Syllabus
Website http://faculty.washington.edu/matsueda/courses/529/web529s16.htm
Time
& Location Thursday 5:30-7:20pm Savery
409
Instructors Ross L. Matsueda
Email matsueda@uw.edu
Office
Hours Mon, 2-3pm,
227 Savery Hall
(&
Thu 7:30-8pm
409
Savery Hall)
PREREQUISITES:
Students
should have a sound background in intermediate statistics for social
scientists, including a basic
course in
statistical inference and the general linear model or multiple regression, as
presented in Sociology
506. Also desirable is a knowledge of (or facility
to learn independently) elementary tools of matrix algebra.
COURSE REQUIREMENTS:
Students will be expected to complete semi-weekly exercises. Several computer assignments will allow
students to analyze
data provided by the instructor and write a brief (no more than 5 pages)
report.
Students will
complete a short (10 pages of text) seminar paper using the methods presented
in the course
using data provided
by the instructor or data from the student.
Alternatively, students may opt to take a final
exam. The paper
will be due on Thur June 7, 5pm. A paper proposal along with a path diagram
will be due
Thur, May 3 in lecture.
All
assignments must be completed on time. A
grade of incomplete will not be given except under
unusual circumstances, such as a family emergency.
GRADING:
Grades will be
based on homework assignments, the seminar paper (or final exam), and possibly
an
unannounced (pop)
exam.
Lecture
Notes: Introduction
to the Course
Lecture
1: Bivariate Linear Model
Lecture
2: Recursive Models & Decomposing Effects
Lab
Notes: Importing an SPSS File into
LISREL
Lecture
3: A Structural Model with Unobservables
Lecture
4: The LISREL SEM Model
Lecture
5: LISREL & PRELIS Programs
Lecture
12: Maximum Likelihood
Lecture
13: Fit Statistics and Multiple Group Models
Lecture
14: WLS and Models for Ordinal Data
Lecture
8: Instrumental Variables & Nonrecursive Models
Lecture
9: MIMIC Models & Identification
Lecture
10: Panel Models & Sibling Models
Lecture
11: Latent Growth Curve Models
Bonus
Lecture: Trajectory Model Diagrams
Importing an
SPSS file into PRELIS
This
memo describes how to import an SPSS save file into PRELIS and then create a
PRELIS system file (*.psf) in LISREL 8.8 (or
equivalently, create a LISREL system file (*.lsf) in
LISREL 9.3). The system file can then be read directly into PRELIS and LISREL.
This
memo describes how to input data (including raw data in ascii format with blanks separating variables, raw
data in ascii format with commas separating
variables, and an SPSS save file. Use
the files, Informal.dat, Informal.csv, and Informal.sav
to practice reading in data into PRELIS and creating a .psf
file (in LISREL 8.8) or a .lsf
file in (LISREL 9.3).
This
memo provides examples of PRELIS runs that create a covariance matrix to input
into LISREL for continuous variable models and a polychoric
correlation matrix and associated asymptotic covariance matrix to be input into
LISREL for ordinal variable models. It
also includes LISREL command files to estimate models
based on the PRELIS saved files. At the
end is annotated output for the two PRELIS runs.
PRELIS Command
Files
This
file reads in a raw data file and computes and saves a
covariance matrix to be input into LISREL (below), which assumes continuous
indicators and normal distributions.
This
file reads raw data into LISREL and computes polychoric correlations and associated asymptotic
covariance matrix. It saves the two
matrices to disk to be input into a LISREL run that
assumes ordinal indicators (see below).
LISREL Command
Files
This
file reads in the covariance matrix from the PRELIS run Preexch1.pr2 (above)
and estimates a one factor confirmatory factor model
using ML under the assumption of continuous and normally distributed
indicators.
This
file reads in the polychoric correlation matrix and associated asymptotic covariance matrix (of the polychoric correlations) and estimates a one-factor
confirmatory factor model using WLS under the assumption of ordinal indicators
and non-normality.
Data Files
This is
an ascii file with blanks
between variables. It consists of four variables taken from the Seattle
Neighborhoods and Crime Survey, which sampled 4,670 residents from 123
neighborhoods (see Matsueda and Drakulich 2016). The
variables are measures of reciprocated exchange, measured on an ordinal scale
1=never, 2=sometimes, 3=often:
How
often have you watched your neighbor's property when they were out of town?
How often have you borrowed tools or
small food items from your neighbors?
How often have you helped a neighbor
with a problem?
How often have you asked neighbors
about personal things like child rearing or jobs?
This is
an ascii file containing raw
data with blanks between variables from the study discussed above. It includes four measures of observed
deviance in the neighborhood (dichotomous measures), plus five measures of
child-centered social control (measured on a four-category ordinal scale (see
Matsueda and Drakulich 2016).
Same
as above, but the ascii file
contains commas between variables.
Same
as above, but this file is an SPSS save file.
Readings:
Matsueda,
Ross L. 2012. “Key Advances in the History of Structural Equation
Modeling.”
Pp. 17-42 in Handbook on Structural Equation Modeling. Edited by Rick H. Hoyle. Guilford Press.
Duncan,
Otis Dudley. 1975. Introduction to
Structural Equation Models. New
York: Academic Press,
Chapters 1 & 2.
Bielby, William T., and Robert M. Hauser. 1977. Introduction to Structural Equation Models. New York:
Academic Press, Chapter 1.
Duncan,
Otis Dudley. 1975. Introduction to
Structural Equation Models. New
York: Academic Press,
Chapters 3 & 4.
Alwin, Duane F. and Robert M. Hauser. 1975. "The
Decomposition of Effects in Path Analysis.” American
Sociological Review 40:37-47.
Paxton,
Pamela, John R. Hipp, and Sandra Marquart-Pyatt.
2011. Nonrecursive Models: Endogeneity,
Reciprocal Relationships, and Feedback Loops. Beverly
Hills: Sage Publications. (Chapter 2.
“Specification in Simultaneous Equation
Models,” pages 4-22.)
Bielby, William T., and Ross L. Matsueda. 1991.
"Statistical Power in Non-Recursive Linear Models." In
Sociological Methodology 1991, Vol. 20,
edited by P. Marsden. Oxford: Basil Blackwell.
John
Robert Warren. 2009. “Socioeconomic Status and Health across the Life
Course: A Test of the Social
Causation and Health Selection Hypotheses.” Social Forces 87: 2125-2154.
Bielby, William T., Robert M. Hauser, and David L. Featherman. 1977. “Response Errors of Black and
Nonblack Males in Models of the
Intergenerational Transmission of Socioeconomic Status.” American
Journal of Sociology 82:1242-88.
Matsueda,
Ross L. 1982. “Testing Control Theory and Differential Association: A
Causal Modeling
Approach. American Sociological Review.” American Sociological Review 47:489-504.
Paxton,
Pamela. 1999. “Is Social Capital Declining in the United States? A Multiple
Indicator
Assessment.” American Journal of
Sociology 105:88-127.
Bartusch,
Dawn Jeglum, and Ross L. Matsueda. 1996. “Gender,
Reflected Appraisals, and Labeling: A
Cross-Group Test of an Interactionist Theory
of Delinquency.” Social Forces 75:145-176.
Lei,
Pui-Wa Lei, and Qiong Wu.
2012. “Estimation in Structural Equation Modeling.” Pp. 164-180 in
Handbook on Structural Equation Modeling. Edited by
Rick H. Hoyle. Guilford Press.
Matsueda,
Ross L., and Kathleen Anderson. 1998. “The Dynamics of Delinquent Peers and
Delinquency.”
Criminology 36:269-308.
Paxton,
Pamela. 2002. “Social Capital and Democracy: An Interdependent
Relationship.” American
Sociological Review 67:254-277.
Hauser,
Robert M., and Peter A. Mossel. 1985. “Fraternal
Resemblance in Educational Attainment and
Occupational Status.” American
Journal of Sociology 91:650-673.
Matsueda,
Ross L. and William T. Bielby. 1986.
"Statistical Power in Covariance Structure Models." Pp.
120-58 in Sociological
Methodology 1986, edited by N.B. Tuma.
Washington, D.C.: American
Sociological Association.