cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 
Try the Materials Informatics Toolkit, which is designed to easily handle SMILES data. This and other helpful add-ins are available in the JMP® Marketplace
Choose Language Hide Translation Bar
madhu
Level III

Latent variables in factor analysis

I intend to collect survey data using structured questionnaire and apply factor analysis and structural equation modelling using JMP Pro. I understand that a latent construct (factor) is traditionally validated by a number of manifest variables. However, I want to add a number of  sub-latent variables for each latent variable. Can it be done in JMP Pro?

Please see below the latent, sub-latent and manifest variables

madhu_0-1735394391570.png

 

madhu_1-1735394406955.png

 

 

5 REPLIES 5
madhu
Level III

Re: Latent variables in factor analysis

Can anyone please suggest me on my above question about factor analysis?

statman
Super User

Re: Latent variables in factor analysis

I'm tagging @LauraCS

"All models are wrong, some are useful" G.E.P. Box
LauraCS
Staff

Re: Latent variables in factor analysis

Hi @madhu,

 

Yes, this is definitely doable in the SEM platform of JMP Pro! Because you already know how to define each of your latent variables (sub-latent and latent--also known as first-order and second-order latent variables), you simply have to use the platform to specify the model you portrayed in your image.

 

Here are some simple steps to accomplish what you need (I'm using the "Job Satisfaction.jmp" data table for illustration):

 

1) Launch the SEM platform with all your manifest variables

     1.a) Open your data table, go to the Analyze menu, and click "Structural Equation Models" under the Multivariate submenu

     1.b) Select all manifest variables in the far-left list, click "Model Variables" and then click "OK"

     

LauraCS_0-1736179134132.png

 

2) Specify your first-order (sub-latent) latent variables

     2.a) Use the "To List" to select the manifest variables that are indicators of the first latent variable, then customize its name by replacing "Latent 1" in the box with the label you prefer, and finally click the "+" button to the right

 

LauraCS_1-1736179525602.png

 

     2.b) Repeat step 2.a until all first-order latent variables are specified (here, I have 3 latent variables labeled: Leadership, Conflict, and Satisfaction)

LauraCS_2-1736179801043.png

 

3) Specify the second-order latent variable

     3.a) The "To List" now shows the names of the first-order latent variables. Use the "To List" to select these latent variables and specify a name for the second-order latent variable using the box under the "To List" and finally click "+" to add the latent variable

LauraCS_3-1736180071947.png

 

The result from the previous steps will be reflected in a path diagram that shows an image very similar to the one supplied in the previous post,

 

LauraCS_4-1736180225298.png

 

Please let me know if I can provide any additional guidance!

~Laura

 

 

Laura C-S
LauraCS
Staff

Re: Latent variables in factor analysis

I should clarify that one can customize the name of the model using the top-left box in the UI. Lastly, clicking the "Run" button will fit the higher-order confirmatory factor analysis model.

 

LauraCS_6-1736180537224.png

 

 

Laura C-S
LauraCS
Staff

Re: Latent variables in factor analysis

@madhu,

 

I should also highlight some of the most useful functionality for these models. After fitting the model, in the Red Triangle Menu, you'll find the option for "Standardized Parameter Estimates." This will display a new table in the report. You can also display these estimates in the path diagram by right-clicking on the blank canvas of the diagram and selecting Show... > Show Estimates > Standardized

 

LauraCS_0-1736181186345.png

 

 

You'll also find "Assess Measurement Model" which will provide you will a dashboard-type report with lots of useful statistics for quantifying the reliability of indicators and constructs, as well as evidence for construct validity. Please see the online documentation for more details on this feature: https://www.jmp.com/support/help/en/18.1/index.shtml#page/jmp/example-of-the-assess-measurement-mode... 

 

HTH,

~Laura

Laura C-S