version 0.4-0

cran · Mar 28, 2020 · f1c02d1 · f1c02d1
1 parent 6ca9327
commit f1c02d1
Show file tree

Hide file tree

Showing 22 changed files with 779 additions and 382 deletions.
diff --git a/DESCRIPTION b/DESCRIPTION
@@ -1,6 +1,6 @@
 Package: MPTmultiverse
 Title: Multiverse Analysis of Multinomial Processing Tree Models
-Version: 0.3-3
+Version: 0.4-0
 Description: 
     Statistical or cognitive modeling usually requires a number of more or less 
     arbitrary choices creating one specific path through a 'garden of forking paths'. 
@@ -22,27 +22,28 @@ Authors@R: c(
     person("Henrik", "Singmann", role = c("aut", "cre"), 
         email="singmann@gmail.com", comment=c(ORCID="0000-0002-4842-3657"))
     , person(c("Daniel", "W."), "Heck", email = "heck@uni-mannheim.de", role = c("aut"))
-    , person("Marius", "Barth", email = "marius.barth@uni-koeln.de", role = c("aut"))
+    , person("Marius", "Barth", email = "marius.barth@uni-koeln.de", role = c("aut"),
+        comment = c(ORCID = "0000-0002-3421-6665"))
     , person("Frederik", "Aust", email = "frederik.aust@uni-koeln.de", role = c("ctb"), 
         comment = c(ORCID = "0000-0003-4900-788X"))
   )
 URL: https://github.com/mpt-network/MPTmultiverse
 BugReports: https://github.com/mpt-network/MPTmultiverse/issues
 Depends: R (>= 2.11.1),
 Imports: parallel, magrittr, tidyr, dplyr, tibble, rlang, reshape2,
-        ggplot2, MPTinR, TreeBUGS, runjags, coda, purrr, readr,
-        limSolve, utils
+        ggplot2, MPTinR, TreeBUGS (>= 1.4.4), runjags, coda, purrr,
+        readr, limSolve, utils
 Suggests: knitr, rmarkdown, testthat
 LazyData: yes
 VignetteBuilder: knitr
-RoxygenNote: 6.1.1
+RoxygenNote: 7.1.0
 License: GPL-2
 NeedsCompilation: no
-Packaged: 2019-08-27 18:50:43 UTC; henrik
+Packaged: 2020-03-27 09:14:02 UTC; henrik
 Author: Henrik Singmann [aut, cre] (<https://orcid.org/0000-0002-4842-3657>),
   Daniel W. Heck [aut],
-  Marius Barth [aut],
+  Marius Barth [aut] (<https://orcid.org/0000-0002-3421-6665>),
   Frederik Aust [ctb] (<https://orcid.org/0000-0003-4900-788X>)
 Maintainer: Henrik Singmann <singmann@gmail.com>
 Repository: CRAN
-Date/Publication: 2019-08-27 23:30:29 UTC
+Date/Publication: 2020-03-28 01:20:02 UTC
diff --git a/MD5 b/MD5
@@ -1,46 +1,46 @@
-77a85bb84b0325bbfeaa0d2abd7bf3b1 *DESCRIPTION
+e58771c2a4e0a4846830e1c61968e39a *DESCRIPTION
 843cf2d97f2f95cdd3b7e728fc64fc68 *NAMESPACE
-d939cb647fa698aa12adc75a499907f1 *R/check_results.R
-91c72a08fdee06ace9b059741d90413a *R/fit_mpt.R
+c43f45d9c8234850f79725c3a9939d6d *NEWS.md
+3dcca1b78e32166996c3efd7f59450ce *R/check_results.R
+3a65c69a07f7c93481919db6bc1e1ddf *R/fit_mpt.R
 a2de0d9664ddbf8e4fa1e7af30b05f3b *R/get_eqn_categories.R
 6a0f6bab5550eb15b4b0395fa899042b *R/get_info.R
-d8604ad9aa9ca4ecfbb5af1f6747c42a *R/make_results_row.R
+28c3dce82255898adca8c906b0f12ffa *R/make_results_row.R
 d75725ae664ca9b0860d575356c7e9fe *R/mpt_options.R
-57177037d82936fd63d1fff358cc0ea4 *R/mptinr.R
+7aa86523c717d54e674c1aa32bff172c *R/mptinr.R
 c0860dc3bb02743f1ff358a5c9afed03 *R/onload.R
 66b4d99a3e9ec352045eb182827c5766 *R/plot_multiverseMPT.R
 faf7d231083e8038c413991ca81d3b2c *R/prep_data_fitting.R
-946377fa7e32a9810e3d558556a8ccac *R/treebugs.R
+95a46328541e1835c8f31faba95d6f8f *R/treebugs.R
 7ba90e1c6fef8537e2ed9d137a801f96 *R/write_results.R
-cf1d73fdd11829c6652ca98c8def0c50 *README.md
-ca94b68a17030b6e8eaa2d868bfaea95 *build/vignette.rds
-5ddc745fc0e241ffe1a2801a5fdf854a *inst/doc/introduction-bayen_kuhlmann_2011.R
-4b71a5c606597665434286519a504557 *inst/doc/introduction-bayen_kuhlmann_2011.html
-8fb0a3da73fb660d23e6e7fb0b8f1dd8 *inst/doc/introduction-bayen_kuhlmann_2011.rmd
+880346b0d7e8d0beb7da9f347de55d93 *README.md
+5fe403db3e748972bb3219944b86e8f4 *build/vignette.rds
+7bf68aaa17ed61961f91891dd914b15f *inst/doc/introduction-bayen_kuhlmann_2011.R
+cf5242568d1f18c48daba9f803b40ba6 *inst/doc/introduction-bayen_kuhlmann_2011.html
+a139dc06efc8599d9ab6b0c7f2c88d92 *inst/doc/introduction-bayen_kuhlmann_2011.rmd
 378c931503b45c257da3b87e4c5e7490 *inst/extdata/2HTSM_Submodel4.eqn
 31de4fabf5cb90e987b2d418a49fed02 *inst/extdata/Kuhlmann_dl7.csv
 d9ebea39a1adb10102dbe3cc044395dc *inst/extdata/prospective_memory.eqn
 0493e5bef231f7a55d9f2cd2a185d011 *inst/extdata/prospective_memory_example.rda
-3acb8d138734233c096e54e6df2c8aa5 *inst/extdata/results_bayen_kuhlmann.RData
+16cb443156825bf6684a7a3bc9ce0dcd *inst/extdata/results_bayen_kuhlmann.RData
 89b7767efbae5229542389ae6bd07922 *inst/extdata/smith_et_al_2011.csv
 b9b0e597e96a6905eb91b6f644f74b95 *man/check_results.Rd
-0697cba09eacca908321aaadacbb044b *man/fit_mpt.Rd
-78ad955cff0898b16a938ba99927ce61 *man/get_info.Rd
+cdb08b7abf35c185d7fed97a48e27b79 *man/fit_mpt.Rd
+97423fc2f954472b92d61b707ce62ae3 *man/get_info.Rd
 9a3023bfc66e5b420f7df718499fb2fa *man/get_pb_output.Rd
-8f3811a251eb5a19d2bbfeea686f1d85 *man/make_results_row.Rd
 91e20cb185f30a2236eeb41c8108fb7d *man/mpt_options.Rd
 70d444459d24aac210c2b85acb17d520 *man/plot.multiverseMPT.Rd
 65b8904cc88c4eea226b24f46b9be466 *man/write_results.Rd
 dd4bfb25654cdb367a4c4485a712e332 *tests/testthat.R
-910e290d3cec7cc844f69da4a542eb04 *tests/testthat/test-data-structures.R
-16a5cbce2a4d7eceb5a9ba2d0cdac561 *tests/testthat/test-identifiability_checks.R
+c04b39aa866ac729df2bf88ddf48c8f0 *tests/testthat/test-data-structures.R
+a60c7e95ba346696712f855ff9673d83 *tests/testthat/test-identifiability_checks.R
 e72d9c5908546dac0b22cc760890a110 *tests/testthat/test-mpt_options.R
-5ffa69e5d65eafdb821e38afd37c70e3 *tests/testthat/test-mptinr.R
-15b9a91654c2a73e11ff9b74ca2b83e8 *tests/testthat/test_treebugs.R
+25031aa67bff49b704a2e2a63919b4ce *tests/testthat/test-mptinr.R
+a5cb9755b915c3c67004fe4301df96f5 *tests/testthat/test_treebugs.R
 7e3b16937fce40bab6396676351497c7 *vignettes/NOT_USED/1_bayen_kuhlmann_2011.html
 378c931503b45c257da3b87e4c5e7490 *vignettes/NOT_USED/2HTSM_Submodel4.eqn
 dd5449a91eba16200d402dbf928414c2 *vignettes/NOT_USED/2_Smith_et_al_2011.rmd
 31de4fabf5cb90e987b2d418a49fed02 *vignettes/NOT_USED/Kuhlmann_dl7.csv
 d9ebea39a1adb10102dbe3cc044395dc *vignettes/NOT_USED/prospective_memory.eqn
 89b7767efbae5229542389ae6bd07922 *vignettes/NOT_USED/smith_et_al_2011.csv
-8fb0a3da73fb660d23e6e7fb0b8f1dd8 *vignettes/introduction-bayen_kuhlmann_2011.rmd
+a139dc06efc8599d9ab6b0c7f2c88d92 *vignettes/introduction-bayen_kuhlmann_2011.rmd
diff --git a/NEWS.md b/NEWS.md
@@ -0,0 +1,16 @@
+
+# MPTmultiverse 0.4-0
+
+- Added a `NEWS.md` file to track changes to the package.
+- Added within-subjects comparisons of parameter estimates that are stored in column
+    `test_within` of the results object.
+    - For the Bayesian methods, these comparisons rely on posterior distributions
+        of parameter differences.
+    - For the no-pooling maximum-likelihood methods, comparisons rely on paired *t* tests.
+    - For the complete-pooling maximum-likelihood method, comparisons rely on the the point estimates
+       of parameters and the estimates of the standard errors of differences calculated
+       from the Hessian matrix.
+- Changed the default behavior for auto-extending Bayesian models: As of now, chains
+    are only combined if Rhat <= Rhat_max*2-1. An informative message is printed whether chains were
+    combined with earlier draws or not.
+- Prepared `MPTmultiverse` for new `tibble` version.
diff --git a/R/check_results.R b/R/check_results.R
@@ -64,17 +64,17 @@ check_results <- function(results) {
 
       not_id <- results %>%
         dplyr::filter(.data$package == "MPTinR" & .data$pooling == "no" & .data$method == meth) %>%
-        dplyr::select("est_indiv") %>%
-        tidyr::unnest() %>%
+        dplyr::select(.data$est_indiv) %>%
+        tidyr::unnest(.data$est_indiv) %>%
         dplyr::group_by(.data$condition, .data$core) %>%
         dplyr::summarise(proportion = mean(!.data$identifiable |
                                              is.na(.data$identifiable))) %>%
         dplyr::ungroup()
 
       not_id2 <- results %>%
         dplyr::filter(.data$package == "MPTinR" & .data$pooling == "no" & .data$method == meth) %>%
-        dplyr::select("est_indiv") %>%
-        tidyr::unnest() %>%
+        dplyr::select(.data$est_indiv) %>%
+        tidyr::unnest(.data$est_indiv) %>%
         dplyr::filter(!.data$identifiable) %>%
         dplyr::group_by(.data$condition, .data$core, .data$parameter) %>%
         dplyr::count() %>%
@@ -105,8 +105,8 @@ check_results <- function(results) {
   tryCatch({
     conv_mptinr_comp <- results %>%
       dplyr::filter(.data$package == "MPTinR" & .data$pooling == "complete") %>%
-      dplyr::select("convergence") %>%
-      tidyr::unnest()
+      dplyr::select(.data$convergence) %>%
+      tidyr::unnest(.data$convergence)
 
     comp_prob <- (conv_mptinr_comp$convergence != 0) |
       (conv_mptinr_comp$rank.fisher != conv_mptinr_comp$n.parameters)
@@ -128,13 +128,13 @@ check_results <- function(results) {
   ### TreeBUGS
   res_tree <- results %>%
     dplyr::filter(.data$package == "TreeBUGS") %>%
-    dplyr::select(!!c("model", "dataset", "pooling", "package", "method", "convergence", "est_group"))
+    dplyr::select(.data$model, .data$dataset, .data$pooling, .data$package, .data$method, .data$convergence, .data$est_group)
 
   for (i in seq_len(nrow(res_tree))) {
     cat("## ", paste(res_tree[i, 1:5], collapse = " // "), ":\n", sep = "")
 
     params <- res_tree[i,] %>%
-      tidyr::unnest(.data$est_group) %>%
+      tidyr::unnest(cols = .data$est_group) %>%
       dplyr::select(.data$parameter, .data$core)
 
     tmp_convergence <- res_tree[i, ]$convergence[[1]] %>%

diff --git a/R/fit_mpt.R b/R/fit_mpt.R
@@ -62,12 +62,9 @@
 #'   the data has more than one between-subjects condition, these need to be
 #'   combined into one condition for this function.
 #'
-#'   Parameter differences or other support for within-subject conditions is not
-#'   provided. The best course of action for within-subjects conditions is to
-#'   simply include separate trees and separate sets of parameters for each
-#'   within-subjects condition. This allows to at least compare the estimates
-#'   for each within-subjects condition across levels of pooling and estimation
-#'   methods.
+#'   To include multiple within-subjects conditions, include separate trees and
+#'   separate sets of parameters for each within-subjects condition in your
+#'   .eqn file.
 #'
 #'   \subsection{Pooling}{
 #'   The following pooling levels are provided (not all by all estimation approaches, see below).
@@ -128,8 +125,11 @@
 #'     For the \emph{complete pooling asymptotic approach}, the group-level parameter
 #'     estimates and goodness-of-fit statistics are the maximum-likelihood and
 #'     G-squared values returned by \code{MPTinR}. The parameter differences are
-#'     based on these values, the standard errors of the difference is simply
-#'     the pooled standard error of the individual parameters. The overall fit
+#'     based on these values. for between-subjects comparisons, the standard
+#'     errors of the differences are simply the pooled standard error of the
+#'     individual parameters; for within-subjects comparisons, the standard errors
+#'     of the differences are based on the respective linear transform of the estimated
+#'     variance-covariance matrix calculated from the Hessian matrix. The overall fit
 #'     (column \code{gof}) is based on an additional fit to the completely
 #'     aggregated data.
 #'
@@ -153,7 +153,9 @@
 #'     statistics are based on a t-test comparing the individual-level estimates
 #'     (again, after excluding non-identifiable estimates). The CIs of the
 #'     difference are based on the SEs (which are derived from a linear model
-#'     equivalent to the t-test).
+#'     equivalent to the t-test). Within-subjects comparisons are based on t-tests
+#'     for paired observations.
+#'
 #'
 #'     The individual-level estimates of the \code{bootstrap based no-pooling}
 #'     approaches are identical to the asymptotic ones. However, the SE is the
@@ -167,9 +169,9 @@
 #'     \code{mpt_options()$max_ci_indiv}, which defaults to \code{0.99}. Thus,
 #'     in the default settings we say a parameter is non-identifiable if the
 #'     bootstrap based CI extends from 0 to 1. The group-level estimates are the
-#'     mean of the identifiable individual-level estimates. And difference
-#'     between conditions is calculated in the same manner as for the asymptotic
-#'     case using the identifiable individual-level parameter esatimates.
+#'     mean of the identifiable individual-level estimates. The difference
+#'     between conditions (as well as within conditions) is calculated in the same manner as for the asymptotic
+#'     case using the identifiable individual-level parameter estimates.
 #'   }
 #'
 #'   \subsection{Bayesian Methods}{
@@ -194,7 +196,7 @@
 #'     For all Bayesian methods, the posterior distribution of the parameters is
 #'     summarized by the posterior mean (in the column \code{est}), posterior
 #'     standard deviation (\code{se}), and credbility intervals (\code{ci_*}).
-#'     For parameter differences (\code{test_between}) and correlations
+#'     For parameter differences (\code{test_between} and \code{test_within}) and correlations
 #'     (\code{est_rho}), Bayesian p-values are computed (column \code{p}) by
 #'     counting the relative proportion of posterior samples that are smaller
 #'     than zero. Goodness of fit is tested with the T1 statistic
@@ -224,6 +226,7 @@
 #'   on the probit scale (only in \code{method="trait"}).
 #'   \item \code{test_between}: Parameter differences between the levels of the
 #'   between-subjects condition (if specified).
+#'   \item \code{test_within}: Within-subjects parameter differences.
 #'   \item \code{gof}: Overall goodness of fit across all individuals.
 #'   \item \code{gof_group}: Group-level goodness of fit.
 #'   \item \code{gof_indiv}: Individual-level goodness of fit.
@@ -326,6 +329,10 @@ fit_mpt <- function(
   data$ExpCond <- as.character(data[[condition]])
   data$Subject <- as.character(data[[id]])
 
+  if(any(duplicated(data$Subject))) {
+    stop("Multiple rows per subject in data. Ensure that the subject identifier is properly specified and you correctly aggregated your data.")
+  }
+
 
   # check MPT file
   mpt_model <- TreeBUGS::readEQN(model)