search

FBartos / RoBMA

RoBMA R package for estimating robust Bayesian meta-analyses
https://fbartos.github.io/RoBMA/
8 stars 3 forks source link

from version 1.2.1 to 2.0.0.6 #14

Closed tressoldi closed 3 years ago

tressoldi commented 3 years ago

I need to run an older syntax with the new version, but I'm not sure about the "translation". May you rewrite the following syntax from v 1.2.1 to v.2.0.0.6?

v. 1.2.1

fit <- RoBMA(y = data$ESg, se = data$SEg, study_names = data$Study, priors_mu = prior("normal", parameters = list(mean = .15, sd = .10)),
priors_tau = prior("invgamma", parameters = list(shape = 1, scale = .15)), priors_omega_null = prior("spike", parameters = list(location = 1)), test_type = "one.sample", priors_mu_null = NULL, priors_tau_null = NULL, priors_omega = NULL, control = list(silent = TRUE), seed = 666)

Thanks Patrizio

FBartos commented 3 years ago

Hi Patrizio.

you probably won't be able to reproduce the results exactly since there were some changes in handling transformations and settings seeds, but this should get you as close as possible I think. Otherwise, the README page describes how to install the old version of the package in order to reproduce your old analyses.

fit <-  RoBMA(
  y = data$ESg, se = data$SEg, study_names = data$Study,
  priors_effect         = prior("normal",   parameters = list(mean = .15, sd = .10)),
  priors_heterogeneity  = prior("invgamma", parameters = list(shape = 1,  scale = .15)),
  priors_bias           = NULL,
  priors_effect_null        = NULL,
  priors_heterogeneity_null = NULL,
  priors_bias_null          = prior_none()
)

(which corresponds to a single model for random effect meta-analyses).

Cheers, Frantisek

tressoldi commented 3 years ago

It reproduces v.1.2.1 exactly. Thank you very much. Patrizio