NA in Std.Err and the CIs, No standard error calculated

[maryam]

Hi,

Yesterday I made a post about whether I could first scale (z-score) the data (because variables have different scales/metric) and also use the argument standardized = TRUE in the summary(). Terrence and PD commented that it is a wrong method and that I should give original raw data and use standardized = TRUE and check the standardized coefficients. When I did so, the Std.Err column, as well as its ci columns give NA and cannot calculate the standard errors; BUT all other values and estimates are EXACTLY the same as the previous code (where I used both scaled data as input and also standardidized= TRUE). Should I ignore the NA and just check the Std.all column? Is there any better way to get rid of below warning messages? Many thanks:

Note: this warning messages only appear if I only give raw data as input and use the standardized argument in summary(); and it disappears when I use the wrong method of scaling the data and using the standardized argument.

Warning messages:
1: In lav_data_full(data = data, group = group, cluster = cluster,  :
  lavaan WARNING: some observed variances are (at least) a factor 1000 times larger than others; use varTable(fit) to investigate
2: In lav_model_vcov(lavmodel = lavmodel, lavsamplestats = lavsamplestats,  :
  lavaan WARNING:
    Could not compute standard errors! The information matrix could
    not be inverted. This may be a symptom that the model is not
    identified.
3: In lav_test_satorra_bentler(lavobject = NULL, lavsamplestats = lavsamplestats,  :
  lavaan WARNING: could not invert information matrix

4: In lav_object_post_check(object) :
  lavaan WARNING: some estimated ov variances are negative
5: In lavaan::lavaan(model = lex.model, data = lexdata22, sample.nobs = 210,  :
  lavaan WARNING: not all elements of the gradient are (near) zero;
                  the optimizer may not have found a local solution;
                  use lavInspect(fit, "optim.gradient") to investigate

[PD]

Estimated parameters should be the same, standard errors should change. Your model did not arrive at a solution so it could not calculate standard errors ( lavaan WARNING: could not invert information matrix ). 

1: In lav_data_full(data = data, group = group, cluster = cluster,  :
  lavaan WARNING: some observed variances are (at least) a factor 1000 times larger than others; use varTable(fit) to investigate

See this post (https://groups.google.com/forum/#!topic/lavaan/FzShzsXYNco)

[maryam]

Thanks PD for your prompt response,

--I checked the link you sent; the person in the post finally used the transformed version of his/her data; Does it give me a reason to use a transformed version of my data as well? e.g., the same scaling (z-score) to get rid of the warning messages?

--In that post, Terrence also replied that a warning message is because there might be a problem; can we proceed to use the values/ estimates and disregarding the warning messages? (e.g., are warning messages to be treated as error messages?)  This is because I get the exact same values for estimates.

--Do I have to report the standard errors? (e.g., is it a strict criterion in papers/thesis/etc?)

Thank you so much for all your help; really appreciate it,