[R-meta] Wald_tests - are these the same or different model?
James Pustejovsky
jepu@to @end|ng |rom gm@||@com
Sat Aug 28 20:47:02 CEST 2021
Hi Catia,
The first test, with constraints = matrix(c(1,0,0,1),2,2), is equivalent to
using constraints = constrain_zero(1:2). In words, the null hypothesis is
that the mean effect size is equal to zero for the first category and for
the second category. Because this null involves constraining two parameters
to zero, the test has df_num = 2.
The second test, with constraints = constrain_pairwise(c(1,2)), is
equivalent to using constraints = constrain_equal(1:2). In words, the null
hypothesis is that the mean effect size in the first category is equal to
the mean effect size in the second category. Because this null involves one
constraint (that the difference between category means is equal to zero),
the test has df_num = 1.
James
On Fri, Aug 27, 2021 at 10:42 PM Cátia Ferreira De Oliveira <
cmfo500 using york.ac.uk> wrote:
> Hello,
>
> I hope you are well.
> I have been working with robumeta models which I have been following with
> Wald_tests to check for differences between levels for categorical
> variables. Whilst I understand the difference between the Wald_tests used
> below for categorical variables with more than 2 levels, is there any
> difference between these models for categorical variables with two levels?
> I ask mostly because they produce different results.
>
>
>
> *robu.model <- robu(formula = yi ~ -1 + as.factor(condition), data = Data,
> studynum = Study, var.eff.size = vi,
> rho = .8, small = TRUE)*
>
> *Wald_test(lit.seq.complexity, constraints = matrix(c(1,0,0,1),2,2), vcov =
> "CR2", tidy = TRUE)*
>
> test Fstat df_num df_denom p_val sig
> HTZ 1.73 2 6.16 0.254
>
> *Wald_test(lit.seq.complexity, constraints = constrain_pairwise(c(1,2)),
> vcov = "CR2", tidy = TRUE)*
>
> test Fstat df_num df_denom p_val sig
> HTZ 0.881 1 6.45 0.382
>
> Thank you,
>
> Catia
>
> [[alternative HTML version deleted]]
>
> _______________________________________________
> R-sig-meta-analysis mailing list
> R-sig-meta-analysis using r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-meta-analysis
>
[[alternative HTML version deleted]]
More information about the R-sig-meta-analysis
mailing list