I'm working on adding meta-regression to an existing Shiny app, and we currently have a tab which displays the result details using the summary() function. We are finding that the summarised output from the relative effects does not display comparisons to the control treatment in the regression model.
We follow these steps:
Create a model using model <- mtc.model()
Using type = "consistency" for the consistency model
Using type = "regression" for the regression model
Generate samples using results <- mtc.run(model)
Calculate relative effects using relEffects <- relative.effect(results, t1 = "the_Little")
Summarise the relative effects using summary(relEffects)
The empirical mean section and the quantiles section display comparisons to the control for the consistency model, but not for the regression model.
The output looks like this for the consistency model:
...
1. Empirical mean and standard deviation for each variable,
plus standard error of the mean:
Mean SD Naive SE Time-series SE
d.the_Little.the_Butcher 0.69365 2.50416 0.0088535 0.2741715
d.the_Little.the_Dung_named 0.25629 2.84530 0.0100597 0.2826365
d.the_Little.the_Great 0.07341 2.18840 0.0077371 0.2117903
d.the_Little.the_Slit_nosed -0.25412 3.08541 0.0109086 0.2909336
d.the_Little.the_Younger -0.15857 2.91021 0.0102891 0.2758437
sd.d 0.09853 0.05806 0.0002053 0.0009789
...
The output looks like this for the regression model:
...
1. Empirical mean and standard deviation for each variable,
plus standard error of the mean:
Mean SD Naive SE Time-series SE
d.the_Great.the_Butcher 0.15784 2.06798 0.0073114 0.216550
d.the_Great.the_Dung_named 0.34709 1.91189 0.0067596 0.201145
d.the_Great.the_Little 0.01633 2.33742 0.0082640 0.241972
d.the_Great.the_Slit_nosed -0.28278 2.34142 0.0082782 0.238055
d.the_Great.the_Younger -0.02894 1.91362 0.0067657 0.155589
sd.d 0.10029 0.05773 0.0002041 0.000954
B 0.17340 4.07162 0.0143953 0.294604
...
It is noted that the summary does have the control treatment set correctly because it is displayed at the bottom of the output:
Regression on "age", shared coefficients, "the_Little" as control
Hi Gert,
I'm working on adding meta-regression to an existing Shiny app, and we currently have a tab which displays the result details using the
summary()
function. We are finding that the summarised output from the relative effects does not display comparisons to the control treatment in the regression model.We follow these steps:
model <- mtc.model()
type = "consistency"
for the consistency modeltype = "regression"
for the regression modelresults <- mtc.run(model)
relEffects <- relative.effect(results, t1 = "the_Little")
summary(relEffects)
The empirical mean section and the quantiles section display comparisons to the control for the consistency model, but not for the regression model.
The output looks like this for the consistency model:
The output looks like this for the regression model:
It is noted that the summary does have the control treatment set correctly because it is displayed at the bottom of the output: