LISNEW

USING LISREL AND PRELIS ON CSDA'S UNIX CLUSTER

LISREL is most often used for structural equation models. It can also be used for confirmatory factor analysis. CSDA has LISREL version 8.30 and PRELIS version 2.30 available on our UNIX machines "liska", "malthus", and "graunt". We also have this software on one publicly available PC in the Computing/Statistical Core.

Data to be used in LISREL can prepared by using the software “PRELIS”. It reads raw data, and creates a data set that can be used easily with LISREL. There are different options within PRELIS that will create different types of data matrices.

The software for LISREL (lisrel8) and PRELIS (prelis2) are stored in /csda/local/src. If you need help setting your .cshrc file so that you can access the software, please contact CSDA or refer to the document "Services Offered by the Computing/Statistical Core of CSDA". (http://www.albany.edu/csda/c_serv.html)

With version 8 of LISREL, you have the option of writing a traditional LISREL program, or, to use the language called "simplis". Simplis is simple, compared to writing a traditional LISREL program (hence the name).

CSDA owns the documentation for LISREL8, and PRELIS2. You are welcome to come to CSDA to look at it. To work with PRESLIS2, you must also look at the documentation for PRELIS1, which is in the same binder. Examples of programs provided with the software can be found in subdirectories in /csda/local/src/EXAMPLES. SSI (the software company) said that they are working on documentation of 8.3, but it is not currently available. You can also visit their WWW site for more information about new features. http://www.ssicentral.com/lisrel/mainlis.htm

To run the prelis, first prepare the program and data files. From within the directory where the program and data reside, issue the command: prelis

The following screen will appear:

P R E L I S 2.30

Karl G. Joreskog and Dag Sorbom
Available Workspace 67108864 bytes

                                                                This program is published exclusively by
                                                                Scientific Software International, Inc.
                                                                7383 N. Lincoln Avenue, Suite 100
                                                                Lincolnwood, IL 60712-1704, U.
                                                                Phone: (800)247-6113, (847)675-0720, Fax: (847)675-2140
                                                                Copyright by Scientific Software International, Inc., 1981-99.
                                                                Use of this program is subject to the terms specified in the
                                                                Universal Copyright Convention.
                                                                Website: www.ssicentral.com

Input file [INPUT] :

Put in the complete name of the input file (using upper and lower case as necessary), and then the following line will appear:

Output File [OUTPUT]:

Provide the name that you want the output file to be written to. This is the file in which prelis will write information about what happened while it was doing its work. If all went well, prelis will also create an output data set with the name that you specified in your program. Following is an example of a prelis program. Lines that begin with "!" are comments. You can document your programs with comment lines.

! sample.pr2. prelis program.
! see page 3-4
! the "DA" line give info about the data. ni = # of vars, NOBS = #cases
DA NI = 6 NObs= 582

! see page 3-12
! next line tells where raw data will be found
RA FI = sample.data

! see page 3-6
! next line lists variables in the order they appear
! in the raw data
LA v1 v2 v3 v4 v5 v6

! see page 3-14 for weight variable

! see page 3-16 for data types. OR = ordinal, CO = continuous, ...
! next line lists variables that are continuous
CO v1 v2 v3 v4 v5 v6

! see page 3-18 to define missing values
! see 3-20 for log and power transformations
! see 3-22 for recodes
! see 3-24 for SC - select cases
! CL gives value labels
! for example, CL v1 0=no 1=yes

! see page 3-29 for information about OU
! specify output. ma=cm means covariance matrix.
! sm provides the name the new data are to be written into

! One user was getting the error message -
! "Arithmetic Exception (core dumped)." and contacted Scientific Software for help.
! The tech support person mentioned that she should add XM to the OUtput line in PRELIS
! which would gives additional storage space and skips over tests of multivariate
! normality. This solved the problem - she was able to get the matrix.

OU ma=cm sm = sample.cm

To run the lisrel, first prepare the program and data files. From within the directory where the program and data reside, issue the command: lisrel

The following screen will appear:

L I S R E L 8.30

Karl G. Joreskog and Dag Sorbom
Available Workspace 67108864 bytes

Input file [INPUT] :

Put in the complete name of the input file (using upper and lower case as necessary), and then the following line will appear:

Output File [OUTPUT]:

Provide the name that you want the output file to be written to. This is the file in which prelis will write information about what happened while it was doing its work and the output for the analysis you requested. See to documentation for information about how to write programs. A small example (not real-world) of what SIMPLIS language looks like follows.

! the title follows
SAMPLE

! give variable names in order that variables appear
OBSERVED VARIABLES
V1 V2 V3 V4 V5 V6

! tell where data are
COVARIANCE MATRIX FROM FILE SAMPLE.CM

! tell sample size
SAMPLE SIZE 582

! define the equations you want estimated
RELATIONSHIPS
V1 = V3 + V4 + V5
V2 = V1 + V3 + V6

! define correlations you want included
LET THE ERRORS OF V1 AND V2 CORRELATE
END OF PROBLEM

Prepared by Patty Glynn and distributed courtesy of the Center for Social and Demographic Analysis, University at Albany, State University of New York. Revised 1/12/200.