P(>|z|) in lavaan output

1,029 views
Skip to first unread message

Claudia Chu

unread,
May 27, 2017, 6:12:00 PM5/27/17
to lavaan
Hello,

I am new to confirmatory factor analysis and lavaan.

I have a very stupid to ask with regard to interpreting the output of my CFA model:  what does the value P(>|z|) mean? Is that the significance level for the factor loading? 


For example, if I have the following results:

Latent Variables:

                   Estimate  Std.Err  z-value  P(>|z|)   Std.lv  Std.all

  VAC =~                                                               

    L1                1.000                                 0.356      0.518

    L2                2.644    0.406    6.519    0.000    0.942    0.734

    L3                2.808    0.545    5.155    0.000    1.001    0.680

    L4                1.144    0.249    4.590    0.000    0.408    0.443


Does that mean there is a non-significant relationship between the latent variable VAC and the manifest variable L1?

Does it matter if the relationship is non-significant? I know a factor loading >0.3 is preferable, but I am not sure how to interpret the significance of the relationship.



Thank you very much.

kma...@aol.com

unread,
May 28, 2017, 10:07:42 AM5/28/17
to lavaan
Claudia,
I think that your columns are somehow misaligned in the first row.  The 0.356 and 0.518 appear to be the standardized factor loading values.  There should be no p value because you cannot fix a parameter and test hypotheses about it too.  The z test only applies to free (constrained or unconstrained) parameters.

Keith
------------------------
Keith A. Markus
John Jay College of Criminal Justice, CUNY
http://jjcweb.jjay.cuny.edu/kmarkus
Frontiers of Test Validity Theory: Measurement, Causation and Meaning.
http://www.routledge.com/books/details/9781841692203/

Terrence Jorgensen

unread,
May 28, 2017, 10:13:22 AM5/28/17
to lavaan
what does the value P(>|z|) mean? Is that the significance level for the factor loading? 

Yes.  A p value is the probability (P()) of observing a statistic (here, the Wald z) at least as extreme as your observed statistic (so, greater than the absolute value: >|z|)

                   Estimate  Std.Err  z-value  P(>|z|)   Std.lv  Std.all

  VAC =~                                                               

    L1                1.000                                 0.356      0.518


Does that mean there is a non-significant relationship between the latent variable VAC and the manifest variable L1?


No, as Keith pointed out while I was typing this :-)

Terrence D. Jorgensen
Postdoctoral Researcher, Methods and Statistics
Research Institute for Child Development and Education, the University of Amsterdam

Claudia

unread,
May 30, 2017, 2:55:31 PM5/30/17
to lav...@googlegroups.com
Thank you for your clarification! I understand now.

Claudia

--
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/XpAyOXx9JBw/unsubscribe.
To unsubscribe from this group and all its topics, send an email to lavaan+unsubscribe@googlegroups.com.
To post to this group, send email to lav...@googlegroups.com.
Visit this group at https://groups.google.com/group/lavaan.
For more options, visit https://groups.google.com/d/optout.

Reply all
Reply to author
Forward
0 new messages