Dear Michael,
I finally see what you meant by your question. I did not realise we now
had to provide a matrix for the constraints.
If you want to test the overall effect of a variable with three groups,
which constraints would you use? Do you have any good resources on how to
do it?
Should it be c(1,1,1,0,0,0) because you would be contrasting the group
levels and not interested in the interactions?
Group.model<- robu(formula = yi ~ 0 + Group+ Group:Variable, data = Data,
studynum = Study, var.eff.size = vi, rho = .8, small = TRUE)
print(Group.model)
Wald_test(Group.model, constraints = ?, vcov = "CR2")
Best wishes,
Catia
On Tue, 10 Aug 2021 at 14:42, C?tia Ferreira De Oliveira <
cmfo500 at york.ac.uk> wrote:
Dear Michael.
I am not sure, I just ran it as it was presented in the blogpost.
Best wishes,
Catia
On Tue, 10 Aug 2021 at 14:08, Michael Dewey <lists at dewey.myzen.co.uk>
wrote:
Dear C?tia
Is constraints allowed to be a vector rather than a matrix?
Michael
On 10/08/2021 12:39, C?tia Ferreira De Oliveira wrote:
Dear all,
I have tried to run the code in this post:
https://www.jepusto.com/clubsandwich-for-rve-meta-analysis/
*options(width=150)*
*library(robumeta)*
*library(clubSandwich)*
*data(dropoutPrevention)*
*m3_robu <- robu(LOR1 ~ study_design + attrition + group_equivalence +
adjusted*
* + outcome + evaluator_independence*
* + male_pct + white_pct + average_age*
* + implementation_quality + program_site + duration +
service_hrs, *
* data = dropoutPrevention, studynum = studyID,
= varLOR, *
* modelweights = "HIER")*
*print(m3_robu)*
*Wald_test(m3_robu, constraints = 10:12, vcov = "CR2")*
But it gives me the following error when I run the Wald_test:
*Error in if (!inherits(constraints, "matrix") | ncol(constraints) !=
: argument is of length zero*
Does anyone know why this is happening?
Best wishes,
Catia
[[alternative HTML version deleted]]