Tools for Those Who Summarize the Evidence Base
Resources and networking for those who conduct or interpret meta-analyses related to any phenomenon that is gauged in multiple studies.
Hi there,I am using A/Prof Mike Cheung's TSSEM (See:…Continue
Started by Ryan Tang. Last reply by Jisoo Ock Oct 7.
Dear All,I am new to MASEM and I am trying to test mediator effect in meta analysis using MASEM. When I run TSSEM 2 I get a saturated model with df =0 for the target model. I want to see how to…Continue
Started by saneesh edacherian. Last reply by saneesh edacherian Oct 5.
Dear all, (sorry if this request is cross-posted) one aspect in meta-analyis that is not entirely clear to is why some meta-analytical studies use the standardized slopes of one (or more)…Continue
Started by Empi. Last reply by Mike Cheung Oct 4.
Dear Mike and others, I am testing a mediation model. I used random effects TSSEM1 and 2 to test whether there is a significant indirect effect overall among my studies--there is! But I also want to…Continue
Started by Mei Yi Ng. Last reply by Mike Cheung Oct 4.
Loading feed
Comment
Dear Empi,
As I don't use STATA, I am not sure how difficult it is to implement the WLS fit function in STATA.
I have a draft on this topic (as a co-author). If you would like to try the STATA code, please drop me an email.
Best,
Mike
Dear all,
my question is whether there is an established minimum number of studies necessary to conduct a random effects meta analysis?
In the neighboring field of multilevel analaysis, the typical advice is to use at least 20-30 groups for random intercept models. However, my impression is that random effects meta-analytical models - which seem to resemble the logic of twolevel analyses - are often based on relatively few studies, say approx. k=10-15. I couldn't find any simulation study, hence I would be very grateful for suggestions...
Best,
Empi
Dear Empi,
#1: I think so though I have not tested it yet. The standard error with the “stdyx” is based on the delta method, which should be the same as the one with the SEM approach.
#2 Yes.
#3 Regarding synthesizing interaction effects, I would suggest using the raw data rather than the summary statistics. If the scales of the variables are comparable across studies, you may compare whether the slopes of the interaction term are equal across samples. It is a three-way interaction.
Best,
Mike
Dear Mike,
another question I am struggling with is how to best synthesize interaction effects (product terms). Again, I have individual-data from several samples with different operationalisations. The interactions are pretty similar across the different studies, and I would like to run a parameter-based meta analysis.
I am not sure how to compute appropriate effect sizes for the product terms, and it also not clear to me which standard errors should be used (for the unstandardised slope? or the standardised slope?). Do you have any informal suggestions?
Best,
Empi
Dear Mike,
I just read your paper “Comparison of methods for constructing confidence intervals of standardized indirect effects”, and would like to ask 2 questions.
#1: Mplus can be used to request standardized indirect effects with associated standard errors (e.g. using “Model indirect” plus “stdyx” in the output). Are these default-standard errors based on the same approach than the standard errors resulting from the Mplus-code you used?
#2: My aim is to run a parameter-based meta-analysis of indirect effects. Would it suffice to square the standard errors of the indirect effects to get the variance needed to compute the inverse of the variance needed for meta-analysing the standardized indirect effects?
Best,
Empi
Dear Nina,
If the data (scales) are comparable across studies, I would suggest using either a multiple-group SEM (fixed-effects model) or a multilevel SEM (random-effects model) in Mplus.
MASEM or using the moderated mediation as the effect size is also fine. But it would be better to maximize the information by using the raw data.
Best,
Mike
A message from Nina Lasek to all members of Meta-analytic structural equation modeling on Meta-Analysis Resources!
Dear meta-analysers,
I replicated a moderated mediation path model across k=11 samples. Indicators and sample sizes (individual data are available, n from approx. 1000 to approx. 2000) differ. But all associations including the moderation of the mediation are in the expected direction.
Currently, the moderation is calculated as a product term using Mplus. I could also do a multigroup analysis with the moderator as grouping variable. Then i would have e.g. two or three groups per sample.
Question is whether such models - either a path model with moderated indirect effects or a multigroup path model with an indirecte effect - might be meta-analysed as MASEM?
Or, perhaps alternatively, I was also thinking about using the regression weights indicating the mediated respectivey the moderated mediated effects as univariate information for a 'conventional' meta-analysis. What do you think? I would be very grateful for tips with regard to applied examples.
Many thanks in advance,
N.
Thank you for your reply - I'll definitely learn R,
best
Empi
Dear Empi,
There is no grouping variable because the data are summary statistics (correlation matrices). test.cov may include the stacked correlation matrices, e.g.,
1
.2 1
.3 .4 1
1
.3 1
.4 .5 1
The sample sizes are specified by NOBSERVATIONS = 100 200;
By the way, I wonder why you would like to use Mplus to do it. Mplus can only fit the fixed-effects model. Moreover, you have to manually setup the constraints.
In R, you can easily do a random-effects model with one line, e.g.,
random1 <- tssem1(data, n, method="REM", RE.type="Diag")
Regards,
Mike
© 2018 Created by Blair T. Johnson. Powered by
You need to be a member of Meta-analytic structural equation modeling to add comments!