Missing = "ML" vs missing = "FIML"

[Maximiliano Escaffi]

Dear all, 

I hope you are well.

I just have a simple questions about how to request Full Information ML for missing data in Lavaan. The tutorial says that to use Full Information ML, missing = "ML" is required in the function. Particularly, it says that:

If the data contain missing values, the default behavior is listwise deletion. If the missing mechanism is MCAR (missing completely at random) or MAR (missing at random), the lavaan package provides case-wise (or ‘full information’) maximum likelihood estimation. You can turn this feature on, by using the argument missing = "ML" when calling the fitting function. An unrestricted (h1) model will automatically be estimated, so that all common fit indices are available.

However, I've seen people here (and elsewhere) suggesting that FIML is actually requested when doing missing = "FIML". For example, in this question in Research Gate, Holger says that to request FIML estimator, one needs to use missing = "FIML" (https://www.researchgate.net/post/How-to-resolve-an-issue-in-R-with-lavaan-installed-using-the-FIML-function).

Can someone help clarify this issue? Specifically, is FIML requested with missing = "FIML" or "ML" or both? And if there is any difference, what is the difference?

Kind regards,

Maximiliano

[Terrence Jorgensen]

is FIML requested with missing = "FIML" or "ML" or both? 

Either.  They are "alias" references to the same method: calculating the (log-)likelihood for the set of variables observed per row of data.

Terrence D. Jorgensen    (he, him, his)
Assistant Professor, Methods and Statistics
Research Institute for Child Development and Education, the University of Amsterdam
http://www.uva.nl/profile/t.d.jorgensen

[Jeremy Miles]

I once heard Bengt Muthen talk about how ML is inherently full information*, and therefore adding "full information" to ML was redundant - at least according to some people. 

Jeremy

* In, for example, mixed models ML is full information, it would be silly to say FIML

--
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/cd90df8d-23f2-40ff-b2eb-dbfa5336b549n%40googlegroups.com.

[Maximiliano Escaffi]

Hi Terrence,

Thank you very much for the clarification. I thought it might be like this but just needed to confirm.

Kind regards,

Maximiliano

--
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/98Nqk5v8Rhc/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/cd90df8d-23f2-40ff-b2eb-dbfa5336b549n%40googlegroups.com.

--

Maximiliano Escaffi-Schwarz, PhD
Organisational Psychologist

[Maximiliano Escaffi]

Right, makes sense.

Thanks!

[Stas Kolenikov]

There's a degree of fullness. You can estimate polychoric correlations pairwise by maximum likelihood, and form a matrix that is not a FIML estimate of all correlations and all ordinal thresholds.

-- Stas Kolenikov, PhD, PStat (ASA, SSC) 

-- Principal Statistician, NORC @NORCnews

-- Opinions stated in this email are mine only, and do not reflect the position of my employer

-- Social media: @StatStas [ Twitter | mastodon.online ]

-- http://stas.kolenikov.name

To view this discussion on the web visit https://groups.google.com/d/msgid/lavaan/CAMtGSxmSYOpBxK4%3Dy3naQL1Q7ipEJ-tZhWnfv%2B3P2AKCoPC0yA%40mail.gmail.com.