Appendix B The LISREL Software In this section
Appendix B: The LISREL Software In this section we will cover how to run LISREL, software designed to run the covariance structure models featured in Chapters 9 and 10, and also 11. Note that LISREL is one of many computer programs available to run these sorts of models. In addition, there is q SAS PROC CALIS q AMOS q LISCOMP q EQS and many other lesser known programs. I am going to cover the method of LISREL commands that is consistent with the matrix approach of this chapter. This means creating "LISREL Project" statements. Other approaches are available with the program. Mathematical Marketing Slide B. 1 LISREL
Example from Johnson and Wichern p. 380 This example utilizes the difference in daily stock prices for five stocks picked from two industrial sectors. We hypothesize that the overall health of the two sectors is causally antecedent to the stock prices of the observed stock prices. Mathematical Marketing Slide B. 2 LISREL
Path Diagram of Johnson and Wichern Data Chemicals Allied Mathematical Marketing Dupont Petroleum Carbide Exxon Texaco Slide B. 3 LISREL
Title Statement Example from Johnson and Wichern p. 380 Describes the DAta DA NO=100 NI=5 LAbels the variables LA Allied Dupont Carbide Exxon Texaco Describes the MOdel MO NY=5 NE=2 PS=SY, FR KM Input Matrix FRee certain elements 1. 000. 577. 509. 387. 462 1. 000. 599. 389. 322 1. 000. 436. 426 1. 000 FR LY(2, 1) LY(3, 1) LY(5, 2) PS(2, 1) Start values for y MA LY 1 0. 4 0 0 1 0. 4 Start values for MA PS 1 -. 2 1 Start values for ST 1 TE(1) - TE(5) OUtput parameters OU NS AL Mathematical Marketing 1. 000. 523 LISREL Program Statements Slide B. 4 LISREL
An Overview of Control Statements First statement is the title DAta parameters NO NI MA xx n. nn … n. nn … … LA first-var-name Mathematical Marketing Number of observations Number of variables Matrix to be analyzed (= KM, CM, MM) n. nn second-var-name ··· Slide B. 5 LISREL
The MOdel Parameters Statement MOdel parameters NY NX NE NK Number of Y variables Number of X variables Number of Eta variables Number of Ksi variables (All of these default to 0) ma can be any of the 8 LISREL parameter matrices: LY, TE, LX, TD, GA, BE, PH, PS Mathematical Marketing ma = FU SY DI ZE ID , FI FR Slide B. 6 LISREL
Creating Start Values Specify a particular value for various elements Specify a whole matrix filled with values ST x. xx ma(m 1, n 1) ma(m 2, n 2) • • • ma x x ··· ··· ··· x x ··· x ma can be LY, TE, LX, TD, GA, BE, PH or PS Mathematical Marketing Slide B. 7 LISREL
Fixing or Freeing Indivual Elements Priority Sequence: 1. Specification on a FI or FR statement 2. Specification in the MO statement 3. Default for that matrix FI ma(m 1, n 1) ma(m 2, n 2) • • • FR ma(m 1, n 1) ma(m 2, n 2) • • • Mathematical Marketing Slide B. 8 LISREL
The OUtput Parameters Statement OU TV MI NS Mathematical Marketing Slide B. 9 LISREL
Special Cases y = y + x = x + = B + + V( ) = q A confirmatory factor model with all y's q A confirmatory factor model with all x's q A structural equation model with all variables observed q A stuctural equation model with all y's Mathematical Marketing Slide B. 10 LISREL
- Slides: 10