The psychological research literature tends to suggest that...

The psychological research literature tends to suggest that drug use and depression are leading indicators of suicide among teenagers. (Note: Set variance of Suicide $=1$ for model identification purposes). Given the following data set information, create and run a LISRELSIMPLIS program to conduct a second-order factor analysis. ``` Observed Variables: drug1 drug2 drug3 drug4 depress1 depress2 depress3 depress4 Sample Size 200 Correlation Matrix 1.000 0.628 1.000 0.623 0.646 1.000 0.542 0.656 0.626 1.000 0.496 0.557 0.579 0.640 1.000 0.374 0.392 0.425 0.451 0.590 1.000 0.406 0.439 0.446 0.444 0.668 . 488 1.000 Standard Deviations 1.379 1.314 1.288 1.388 1.405 1.269 1.435 1.423 Latent Variables: drugs depress Suicide ``` The second-order factor model is diagrammed in Figure 14.4: ( FIGURE CAN'T COPY )

The psychological research literature tends to suggest that drug use and depression are leading indicators of suicide among teenagers. (Note: Set variance of Suicide $=1$ for model identification purposes). Given the following data set information, create and run a LISRELSIMPLIS program to conduct a second-order factor analysis.
```
Observed Variables: drug1 drug2 drug3 drug4 depress1
depress2 depress3 depress4
Sample Size 200
Correlation Matrix
1.000
0.628 1.000
0.623 0.646 1.000
0.542 0.656 0.626 1.000
0.496 0.557 0.579 0.640 1.000
0.374 0.392 0.425 0.451 0.590 1.000
0.406 0.439 0.446 0.444 0.668 . 488 1.000
Standard Deviations 1.379 1.314 1.288 1.388 1.405 1.269
1.435 1.423
Latent Variables: drugs depress Suicide
```
The second-order factor model is diagrammed in Figure 14.4:
( FIGURE CAN'T COPY )

Key Concepts

Structural Equation Modeling

Structural Equation Modeling (SEM) is a statistical framework that combines aspects of factor analysis and multiple regression. It is used to model complex relationships between observed variables and latent constructs, allowing researchers to assess direct and indirect effects among variables and test theoretical models against empirical data.

Latent and Observed Variables

Latent variables represent unobserved theoretical constructs inferred from measured (observed) variables. Observed variables are the directly measured indicators that reflect underlying latent constructs. This distinction is crucial in models where abstract concepts, such as psychological traits or behaviors, need to be quantified through multiple indicators.

Second-Order Factor Analysis

Second-order factor analysis is an extension of traditional factor analysis where the first-order factors themselves load onto a higher-order (second-order) factor. This approach is often used when multiple dimensions (or first-order factors) are believed to be influenced by a single overarching construct, providing a more parsimonious representation of the data.

Model Identification

Model identification refers to the process of ensuring that there is a unique solution for the parameters being estimated in a statistical model. In SEM, constraints such as fixing the variance of a latent variable (e.g., setting the variance of 'Suicide' to 1) are employed to achieve identification and to facilitate meaningful parameter estimation.

LISREL Programming

LISREL (Linear Structural Relations) is a software package specifically designed for conducting structural equation modeling and factor analysis. It uses a programming language to specify the model, input data, and constraints, enabling researchers to implement complex models, such as second-order factor analyses, and assess their fit to the data.

Transcript

Please give Ace some feedback