文档介绍：Module 6: Structural Equation Models (LISREL)
Sheets for the 4th Week
• Some Notes on Matrix Algebra
• The LISREL Model as a Set of Matrices
• The mand language
• Simultaneous Multiple Group Analysis
• Structured Means Analysis Using LISREL and SIMPLIS Language
• Problems in Structured Means Analysis
• Structured Means Analysis or Variance Analysis (Dummy Regression)
(steffen.******@-
-) 1
Some Notes on Matrix Algebra
A LISREL model explaining efficacy based on ALLBUS 1996 data
(standardized solution)
(.133)
δ
1
Polint1
.205
X1 (.722)
.931 ε
1
POLINT
ξ
1 ε
Polint2 .963 Y3 3 Impact
X2 .083 (.515) Y1
.696
.527
δ TRUST .611 EFFICACY
2 ηη
1 2
(.073) -.037
.726
.266 1 (.473) ζ
Governm 2 Election
(.924) ε
Y4 4 (.532) Y2
Leader 1 LEADER ε
2
X3 ξ
2 .087 (.722)
Minimum Fit Function Chi-Square = (P = )
2
Data from ALLBUS 1996 (n=1882)
Polint1 Polint2 Impact Election Governm Leader
-------- -------- -------- -------- -------- -------- --------
Polint1
Polint2
Impact
Election
Governm
Leader - -
Variables:
Polint1 & Polint2: Political Interest;
Impact : Impact of Citizen on politic;
Election: Elections are efficient methods;
: Politicans keep their promises;
Governm: Government will do best
Leader: Grateful to have leaders who tell us what to do and how to do
3
SIMPLIS Input:
SIMPLIS Model for ALLBUS data with LISREL output
Observed Variables
Polint1 Polint2 Impact Election Governm Leader
Covariance Matrix From File
Sample Size = 1882
Latent Variables POLINT TRUST EFFICACY LEADER
Relationships
Polint1 Polint2 = 1*POLINT
Impact Election =1*EFFICACY
Governm =1*TRUST
Leader =1*LEADER
TRUST =POLI