Dealing with too many zero values of dependent variables

[Md. Shahin]

Dear Colleagues

I have designed a model, where there are two latent variables that predict two dependent variables. Socioeconomic variables are also being used as predictors along with latent variables. My problem is that there are too many zero values for dependent variables. How can I deal with this issue of zero values in SEM modelling? 

How can I interpret the results?

Your help is highly appreciated.  

[Pat Malone]

Look into "semicontinuous" or "two-part" models. You might also have a "hurdle" outcome. Those present a few options, depending in part on why you think you have so many zeroes.

--
You received this message because you are subscribed to the Google Groups "lavaan" group.
To unsubscribe from this group and stop receiving emails from it, send an email to lavaan+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/lavaan/9d39cf66-382b-4ca0-934b-0d2dc2c55e6cn%40googlegroups.com.

--

Patrick S. Malone, PhD

Sr Research Statistician, FAR HARBΦR

+1 803.553.4181  |  pat@  |  farharbor.com

This message may contain confidential information; if you are not the intended recipient please notify the sender and delete the message.

[Pat Malone]

Another search term is "zero-inflated" outcomes.

[Md. Shahin]

Hi Pat

I am new to hurdle modelling. May I do that in lavaan?

Cheers!

Md. Shahin 

Associate Professor
Department of Disaster Resilience and Engineering

Patuakhali Science and Technology University, Bangladesh.
Web site: https://www.pstu.ac.bd/teachers/shahin

Email: shahi...@pstu.ac.bd

;shahinps...@gmail.com

You received this message because you are subscribed to a topic in the Google Groups "lavaan" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/lavaan/nazeqXJAAjM/unsubscribe.
To unsubscribe from this group and all its topics, send an email to lavaan+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/lavaan/CAJhmz4f2SU5xUJ_%3DyqiA%3DzH4W14p-B%3DQxCd1DODQrfGkmmeSNw%40mail.gmail.com.

[Terrence Jorgensen]

I am new to hurdle modelling. May I do that in lavaan?

Not in any automatic way.  There are papers showing how to hack Mplus by making 2 copies of each zero-inflated variable:

• A binary indicator of whether it is nonzero (with NA assigned to truly missing values)
• A continuous indicator of observed nonzero values, which is NA for any observations that are missing or 0

In the Mplus articles, they use marginal MLE to accommodate missingness with FIML, but MML estimation is not currently an option in lavaan.  With minimal (true) NAs, I think you might be able to use WLSMV or PML estimation, but I have not seen those evaluated with simulations, so I hesitate to recommend it.

Terrence D. Jorgensen

Assistant Professor, Methods and Statistics

Research Institute for Child Development and Education, the University of Amsterdam

http://www.uva.nl/profile/t.d.jorgensen

 

[Md. Shahin]

Hi Terrence D. Jorgensen

Thanks for your suggestions. 

Cheers!

Md. Shahin 

Associate Professor
Department of Disaster Resilience and Engineering

Patuakhali Science and Technology University, Bangladesh.
Web site: https://www.pstu.ac.bd/teachers/shahin

Email: shahi...@pstu.ac.bd

;shahinps...@gmail.com

--

You received this message because you are subscribed to a topic in the Google Groups "lavaan" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/lavaan/nazeqXJAAjM/unsubscribe.
To unsubscribe from this group and all its topics, send an email to lavaan+un...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/lavaan/1a92fb46-d7bf-44bf-a6e9-1e8e397706d0n%40googlegroups.com.