Closed IndrajeetPatil closed 3 years ago
Not sure what you expect on the parameters
side? This is the default behavior of correlationBF
/ ttestBF
(different tests use differently scales priors...).
Ach, apologies.
I thought we had opinionated priors for these tests, but now that you mention this, I realize this inconsistency stems from BayesFactor
itself.
@mattansb Thinking of this further, another thing I notice is that Prior_Scale
values are NA
for anovaBF
outputs.
Should we be including rscaleFixed
and rscaleRandom
argument values here?
library(BayesFactor)
#> Loading required package: coda
#> Loading required package: Matrix
#> ************
#> Welcome to BayesFactor 0.9.12-4.2. If you have questions, please contact Richard Morey (richarddmorey@gmail.com).
#>
#> Type BFManual() to open the manual.
#> ************
library(parameters)
data(puzzles)
result = anovaBF(RT ~ shape*color + ID, data = puzzles, whichRandom = "ID",
whichModels = 'top', progress=FALSE)
as.data.frame(model_parameters(result))$Prior_Scale
#> [1] NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA
It's hard to match the different parameters to the rscale*
values (I've had a little back and forth about this with Richard some time ago). That is why, currently, bayestestR::describe_prior()
behaves like this:
describe_prior(result)
#> Parameter Prior_Distribution Prior_Location Prior_Scale
#> 1 fixed cauchy 0 0.5000000
#> 2 random cauchy 0 1.0000000
#> 3 continuous cauchy 0 0.3535534
(And does not return the dist/loc/scale on a parameter basis...)
It is unfortunate that the docs and internals from BayesFactor
are somewhat opaque... ๐คทโโ๏ธ
I suggest opening this as a feature request on bayestestR
- if it works there, it will work here as well.
Doesn't describe_priors rely itself on insight::get_priors
๐
I'm not sure I understand what is the issue?
The issue is that Prior_Scale
column from model_parameters
for anovaBF
contains NA
s, while it is not NA
for the other BF tests.
For example, one-sample t-test
library(BayesFactor)
library(parameters)
model <- ttestBF(x = rnorm(100, 1, 1))
as.data.frame(model_parameters(model))
#> Parameter Median CI_low CI_high pd ROPE_Percentage Prior_Distribution
#> 1 Difference 0.8943024 0.7003134 1.105907 1 0 cauchy
#> Prior_Location Prior_Scale Effects Component BF
#> 1 0 0.7071068 fixed conditional 7.536535e+15
This is what I get:
library(BayesFactor)
#> Loading required package: coda
#> Loading required package: Matrix
#> ************
#> Welcome to BayesFactor 0.9.12-4.2. If you have questions, please contact Richard Morey (richarddmorey@gmail.com).
#>
#> Type BFManual() to open the manual.
#> ************
data(puzzles)
result = anovaBF(RT ~ shape*color + ID, data = puzzles, whichRandom = "ID",
whichModels = 'top', progress=FALSE)
insight::get_priors(result)
#> Parameter Distribution Location Scale
#> 1 fixed cauchy 0 0.5000000
#> 2 random cauchy 0 1.0000000
#> 3 continuous cauchy 0 0.3535534
bayestestR::describe_posterior(result)
#> Multiple `BFBayesFactor` models detected - posteriors are extracted from the first numerator model.
#> See help("get_parameters", package = "insight").
#> Multiple `BFBayesFactor` models detected - posteriors are extracted from the first numerator model.
#> See help("get_parameters", package = "insight").
#> Multiple `BFBayesFactor` models detected - posteriors are extracted from the first numerator model.
#> See help("get_parameters", package = "insight").
#> Multiple `BFBayesFactor` models detected - posteriors are extracted from the first numerator model.
#> See help("get_parameters", package = "insight").
#> Parameter Median CI CI_low CI_high pd ROPE_CI
#> 20 mu 44.9995360 89 43.81886593 46.0112048 1.00000 89
#> 22 shape-round 0.4287515 89 0.14221308 0.7567978 0.98975 89
#> 23 shape-square -0.4287515 89 -0.75679784 -0.1422131 0.98975 89
#> 1 color-color -0.4286762 89 -0.73877251 -0.1367203 0.98750 89
#> 2 color-monochromatic 0.4286762 89 0.13672027 0.7387725 0.98750 89
#> 8 ID-1 2.4739495 89 1.09330337 3.9717513 0.99625 89
#> 12 ID-2 0.4195921 89 -0.94950711 1.8824135 0.69700 89
#> 13 ID-3 0.8798805 89 -0.56285774 2.3080603 0.85300 89
#> 14 ID-4 0.4767006 89 -0.89647667 1.8895565 0.69275 89
#> 15 ID-5 3.1602326 89 1.73420933 4.6631944 1.00000 89
#> 16 ID-6 0.4772122 89 -1.01083371 1.8206981 0.69625 89
#> 17 ID-7 -3.1566652 89 -4.61220177 -1.6353605 0.99975 89
#> 18 ID-8 -0.2518115 89 -1.67235185 1.2315339 0.58625 89
#> 19 ID-9 -2.4922985 89 -3.90178050 -1.0604600 0.99625 89
#> 9 ID-10 0.6498484 89 -0.66506693 2.1864021 0.78450 89
#> 10 ID-11 0.6704275 89 -0.77048153 2.0721755 0.78175 89
#> 11 ID-12 -3.3874473 89 -4.89478836 -1.9327297 0.99975 89
#> 24 sig2 1.7511495 89 1.18317492 2.5900721 1.00000 89
#> 7 g_shape 0.3527367 89 0.01828558 2.2414989 1.00000 89
#> 5 g_color 0.3418209 89 0.02022153 2.2027349 1.00000 89
#> 6 g_ID 2.6587834 89 0.80518350 5.3445872 1.00000 89
#> 3 continuous NA NA NA NA NA NA
#> 4 fixed NA NA NA NA NA NA
#> 21 random NA NA NA NA NA NA
#> ROPE_low ROPE_high ROPE_Percentage BF Prior_Distribution
#> 20 -0.1 0.1 0.00000000 2.6132620 <NA>
#> 22 -0.1 0.1 0.00000000 0.2381810 <NA>
#> 23 -0.1 0.1 0.00000000 0.2362146 <NA>
#> 1 -0.1 0.1 0.00000000 2.6132620 <NA>
#> 2 -0.1 0.1 0.00000000 0.2381810 <NA>
#> 8 -0.1 0.1 0.00000000 0.2362146 <NA>
#> 12 -0.1 0.1 0.07862960 2.6132620 <NA>
#> 13 -0.1 0.1 0.05953384 0.2381810 <NA>
#> 14 -0.1 0.1 0.09126650 0.2362146 <NA>
#> 15 -0.1 0.1 0.00000000 2.6132620 <NA>
#> 16 -0.1 0.1 0.08480764 0.2381810 <NA>
#> 17 -0.1 0.1 0.00000000 0.2362146 <NA>
#> 18 -0.1 0.1 0.10249930 2.6132620 <NA>
#> 19 -0.1 0.1 0.00000000 0.2381810 <NA>
#> 9 -0.1 0.1 0.07722550 0.2362146 <NA>
#> 10 -0.1 0.1 0.06402696 2.6132620 <NA>
#> 11 -0.1 0.1 0.00000000 0.2381810 <NA>
#> 24 -0.1 0.1 0.00000000 0.2362146 <NA>
#> 7 -0.1 0.1 0.11738276 2.6132620 <NA>
#> 5 -0.1 0.1 0.13591688 0.2381810 <NA>
#> 6 -0.1 0.1 0.00000000 0.2362146 <NA>
#> 3 NA NA NA NA cauchy
#> 4 NA NA NA NA cauchy
#> 21 NA NA NA NA cauchy
#> Prior_Location Prior_Scale
#> 20 NA NA
#> 22 NA NA
#> 23 NA NA
#> 1 NA NA
#> 2 NA NA
#> 8 NA NA
#> 12 NA NA
#> 13 NA NA
#> 14 NA NA
#> 15 NA NA
#> 16 NA NA
#> 17 NA NA
#> 18 NA NA
#> 19 NA NA
#> 9 NA NA
#> 10 NA NA
#> 11 NA NA
#> 24 NA NA
#> 7 NA NA
#> 5 NA NA
#> 6 NA NA
#> 3 0 0.3535534
#> 4 0 0.5000000
#> 21 0 1.0000000
parameters::model_parameters(result)
#> Multiple `BFBayesFactor` models detected - posteriors are extracted from the first numerator model.
#> See help("get_parameters", package = "insight").
#> Multiple `BFBayesFactor` models detected - posteriors are extracted from the first numerator model.
#> See help("get_parameters", package = "insight").
#> Multiple `BFBayesFactor` models detected - posteriors are extracted from the first numerator model.
#> See help("get_parameters", package = "insight").
#> Multiple `BFBayesFactor` models detected - posteriors are extracted from the first numerator model.
#> See help("get_parameters", package = "insight").
#> # Extra Parameters
#>
#> Parameter | Median | 89% CI | pd | % in ROPE | BF
#> -------------------------------------------------------------
#> mu | 45.00 | [43.99, 46.21] | 100% | 0% | 2.61
#> sig2 | 1.75 | [ 1.16, 2.51] | 100% | 0% | 0.24
#> g_shape | 0.36 | [ 0.02, 2.35] | 100% | 13.17% | 2.61
#> g_color | 0.35 | [ 0.02, 2.25] | 100% | 12.58% | 0.24
#> g_ID | 2.64 | [ 0.83, 5.29] | 100% | 0% | 0.24
#>
#> # Fixed Effects
#>
#> Parameter | Median | 89% CI | pd | % in ROPE | BF
#> -------------------------------------------------------------------------
#> shape-round | 0.43 | [ 0.11, 0.73] | 98.67% | 0% | 0.24
#> shape-square | -0.43 | [-0.73, -0.11] | 98.67% | 0% | 0.24
#> color-color | -0.43 | [-0.73, -0.12] | 99.12% | 0% | 2.61
#> color-monochromatic | 0.43 | [ 0.12, 0.73] | 99.12% | 0% | 0.24
#>
#> # Random Effects
#>
#> Parameter | Median | 89% CI | pd | % in ROPE | BF
#> ---------------------------------------------------------------
#> ID-1 | 2.49 | [ 0.99, 3.93] | 99.62% | 0% | 0.24
#> ID-2 | 0.45 | [-1.10, 1.82] | 68.20% | 9.35% | 2.61
#> ID-3 | 0.91 | [-0.58, 2.34] | 83.90% | 6.04% | 0.24
#> ID-4 | 0.45 | [-0.96, 1.87] | 69.27% | 8.59% | 0.24
#> ID-5 | 3.18 | [ 1.66, 4.63] | 99.92% | 0% | 2.61
#> ID-6 | 0.43 | [-0.99, 1.85] | 70.33% | 9.46% | 0.24
#> ID-7 | -3.19 | [-4.70, -1.69] | 99.95% | 0% | 0.24
#> ID-8 | -0.23 | [-1.67, 1.25] | 60.58% | 10.28% | 2.61
#> ID-9 | -2.48 | [-3.97, -1.04] | 99.42% | 0% | 0.24
#> ID-10 | 0.66 | [-0.80, 2.10] | 77.72% | 7.89% | 0.24
#> ID-11 | 0.66 | [-0.80, 2.16] | 77.30% | 7.64% | 2.61
#> ID-12 | -3.40 | [-4.84, -1.92] | 99.95% | 0% | 0.24
Created on 2020-09-12 by the reprex package (v0.3.0)
Please re-open an issue if it's not resolved, else the discussion gets lost.
Ah, thanks for the details. I guess then the issue is if the info about prior scale present in bayestestR::describe_posterior
can also be included in parameters::model_parameters
for anovaBF
outputs.
(note that I fixed the weird BF column given here - these should all be the same value for a given model)
I thought the issue was that the location/scale were not given for each parameter (what Indrajeet wanted?), but is given for the different parameter types (fixed, random...).
-- Mattan S. Ben-Shachar, PhD student Department of Psychology & Zlotowski Center for Neuroscience Ben-Gurion University of the Negev The Developmental ERP Lab
On Sat, Sep 12, 2020, 09:47 Indrajeet Patil notifications@github.com wrote:
Ah, thanks for the details. I guess then the issue is if the info about prior scale present in bayestestR::describe_posterior can also be included in parameters::model_parameters.
โ You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/easystats/bayestestR/issues/336#issuecomment-691427675, or unsubscribe https://github.com/notifications/unsubscribe-auth/AINRP6EQCANYPZI6J5N4J7TSFMKRXANCNFSM4RH5SNBA .
If I understand, in order to fix it and to have prior information displayed alongside each parameter, we would in theory need to retrieve which parameters are "continuous" "fixed" and "random" and combine accordingly with the parameters table is that correct?
Yes, that's how I understood this as well.
I feel like it would make sense to retrieve this kind of info at insight's level no?
Should I transfer this issue to insight
repo then?
I think so yes? Because it's not really tied to Bayesian stuff per se but rather an issue of retrieving / tidying some info from a given package ๐คทโโ๏ธ
Now we need someone who knows whether a parameter is fixed, random or continuous.
This won't work for the cases detailed in #223 ...
get_type <- function(trm) {
if (grepl(":", trm, fixed = TRUE)) {
trm <- unlist(strsplit(trm, ":", fixed = TRUE))
}
if (any(trm %in% dataTypes[["random"]])) {
"random"
} else if (any(trm %in% dataTypes[["continuous"]])) {
"continuous"
} else if (any(trm %in% dataTypes[["fixed"]])) {
"fixed"
} else {
""
}
}
library(BayesFactor)
#> Loading required package: coda
#> Loading required package: Matrix
#> ************
#> Welcome to BayesFactor 0.9.12-4.2. If you have questions, please contact Richard Morey (richarddmorey@gmail.com).
#>
#> Type BFManual() to open the manual.
#> ************
iris$ID <- factor(rep(1:30, each = 5))
res <- lmBF(Sepal.Length ~ Species * Sepal.Width + Petal.Width + ID + Sepal.Width:ID, data = iris,
whichRandom = c("ID","Sepal.Width:ID"),
progress = FALSE)
#> Warning in doNwaySampling(method, y, X, rscale, iterations, gMap, incCont, :
#> Some NAs were removed from sampling results: 10000 in total.
pars <- insight::get_parameters(res)
if (insight::model_info(res)$is_linear) {
dataTypes <- res@numerator[[1]]@dataTypes
dataTypes <- tapply(names(dataTypes), dataTypes, "[", simplify = FALSE)
par_names_clean <- sub("\\-(.*)", "", colnames(pars))
# par_names_clean <- par_names_clean[-(which(par_names_clean=="sig2"):length(par_names_clean))]
par_types <- sapply(par_names_clean, get_type)
}
par_types <- setNames(names(par_types),par_types)
priors <- insight::get_priors(res)
idx <- sapply(names(par_types), function(x) {
if (!any(i <- x==priors$Parameter)) return(NA)
which(i)
})
priors <- priors[idx,]
priors$Parameter <- colnames(pars)
priors
#> Parameter Distribution Location
#> NA mu <NA> NA
#> 1 Species-setosa cauchy 0
#> 1.1 Species-versicolor cauchy 0
#> 1.2 Species-virginica cauchy 0
#> 3 Sepal.Width-Sepal.Width cauchy 0
#> 3.1 Petal.Width-Petal.Width cauchy 0
#> 2 ID-1 cauchy 0
#> 2.1 ID-2 cauchy 0
#> 2.2 ID-3 cauchy 0
#> 2.3 ID-4 cauchy 0
#> 2.4 ID-5 cauchy 0
#> 2.5 ID-6 cauchy 0
#> 2.6 ID-7 cauchy 0
#> 2.7 ID-8 cauchy 0
#> 2.8 ID-9 cauchy 0
#> 2.9 ID-10 cauchy 0
#> 2.10 ID-11 cauchy 0
#> 2.11 ID-12 cauchy 0
#> 2.12 ID-13 cauchy 0
#> 2.13 ID-14 cauchy 0
#> 2.14 ID-15 cauchy 0
#> 2.15 ID-16 cauchy 0
#> 2.16 ID-17 cauchy 0
#> 2.17 ID-18 cauchy 0
#> 2.18 ID-19 cauchy 0
#> 2.19 ID-20 cauchy 0
#> 2.20 ID-21 cauchy 0
#> 2.21 ID-22 cauchy 0
#> 2.22 ID-23 cauchy 0
#> 2.23 ID-24 cauchy 0
#> 2.24 ID-25 cauchy 0
#> 2.25 ID-26 cauchy 0
#> 2.26 ID-27 cauchy 0
#> 2.27 ID-28 cauchy 0
#> 2.28 ID-29 cauchy 0
#> 2.29 ID-30 cauchy 0
#> 3.2 Species:Sepal.Width-setosa.&.Sepal.Width cauchy 0
#> 3.3 Species:Sepal.Width-versicolor.&.Sepal.Width cauchy 0
#> 3.4 Species:Sepal.Width-virginica.&.Sepal.Width cauchy 0
#> 2.30 Sepal.Width:ID-1 cauchy 0
#> 2.31 Sepal.Width:ID-2 cauchy 0
#> 2.32 Sepal.Width:ID-3 cauchy 0
#> 2.33 Sepal.Width:ID-4 cauchy 0
#> 2.34 Sepal.Width:ID-5 cauchy 0
#> 2.35 Sepal.Width:ID-6 cauchy 0
#> 2.36 Sepal.Width:ID-7 cauchy 0
#> 2.37 Sepal.Width:ID-8 cauchy 0
#> 2.38 Sepal.Width:ID-9 cauchy 0
#> 2.39 Sepal.Width:ID-10 cauchy 0
#> 2.40 Sepal.Width:ID-11 cauchy 0
#> 2.41 Sepal.Width:ID-12 cauchy 0
#> 2.42 Sepal.Width:ID-13 cauchy 0
#> 2.43 Sepal.Width:ID-14 cauchy 0
#> 2.44 Sepal.Width:ID-15 cauchy 0
#> 2.45 Sepal.Width:ID-16 cauchy 0
#> 2.46 Sepal.Width:ID-17 cauchy 0
#> 2.47 Sepal.Width:ID-18 cauchy 0
#> 2.48 Sepal.Width:ID-19 cauchy 0
#> 2.49 Sepal.Width:ID-20 cauchy 0
#> 2.50 Sepal.Width:ID-21 cauchy 0
#> 2.51 Sepal.Width:ID-22 cauchy 0
#> 2.52 Sepal.Width:ID-23 cauchy 0
#> 2.53 Sepal.Width:ID-24 cauchy 0
#> 2.54 Sepal.Width:ID-25 cauchy 0
#> 2.55 Sepal.Width:ID-26 cauchy 0
#> 2.56 Sepal.Width:ID-27 cauchy 0
#> 2.57 Sepal.Width:ID-28 cauchy 0
#> 2.58 Sepal.Width:ID-29 cauchy 0
#> 2.59 Sepal.Width:ID-30 cauchy 0
#> NA.1 sig2 <NA> NA
#> NA.2 g_Species <NA> NA
#> NA.3 g_ID <NA> NA
#> NA.4 g_continuous <NA> NA
#> Scale
#> NA NA
#> 1 0.5000000
#> 1.1 0.5000000
#> 1.2 0.5000000
#> 3 0.3535534
#> 3.1 0.3535534
#> 2 1.0000000
#> 2.1 1.0000000
#> 2.2 1.0000000
#> 2.3 1.0000000
#> 2.4 1.0000000
#> 2.5 1.0000000
#> 2.6 1.0000000
#> 2.7 1.0000000
#> 2.8 1.0000000
#> 2.9 1.0000000
#> 2.10 1.0000000
#> 2.11 1.0000000
#> 2.12 1.0000000
#> 2.13 1.0000000
#> 2.14 1.0000000
#> 2.15 1.0000000
#> 2.16 1.0000000
#> 2.17 1.0000000
#> 2.18 1.0000000
#> 2.19 1.0000000
#> 2.20 1.0000000
#> 2.21 1.0000000
#> 2.22 1.0000000
#> 2.23 1.0000000
#> 2.24 1.0000000
#> 2.25 1.0000000
#> 2.26 1.0000000
#> 2.27 1.0000000
#> 2.28 1.0000000
#> 2.29 1.0000000
#> 3.2 0.3535534
#> 3.3 0.3535534
#> 3.4 0.3535534
#> 2.30 1.0000000
#> 2.31 1.0000000
#> 2.32 1.0000000
#> 2.33 1.0000000
#> 2.34 1.0000000
#> 2.35 1.0000000
#> 2.36 1.0000000
#> 2.37 1.0000000
#> 2.38 1.0000000
#> 2.39 1.0000000
#> 2.40 1.0000000
#> 2.41 1.0000000
#> 2.42 1.0000000
#> 2.43 1.0000000
#> 2.44 1.0000000
#> 2.45 1.0000000
#> 2.46 1.0000000
#> 2.47 1.0000000
#> 2.48 1.0000000
#> 2.49 1.0000000
#> 2.50 1.0000000
#> 2.51 1.0000000
#> 2.52 1.0000000
#> 2.53 1.0000000
#> 2.54 1.0000000
#> 2.55 1.0000000
#> 2.56 1.0000000
#> 2.57 1.0000000
#> 2.58 1.0000000
#> 2.59 1.0000000
#> NA.1 NA
#> NA.2 NA
#> NA.3 NA
#> NA.4 NA
Created on 2020-09-15 by the reprex package (v0.3.0)
We should first fix clean_parameters()
for such models. This is usually the base for matching parameters:
library(insight)
library(BayesFactor)
#> Loading required package: coda
#> Loading required package: Matrix
#> ************
#> Welcome to BayesFactor 0.9.12-4.2. If you have questions, please contact Richard Morey (richarddmorey@gmail.com).
#>
#> Type BFManual() to open the manual.
#> ************
iris$ID <- factor(rep(1:30, each = 5))
m1 <- lmBF(Sepal.Length ~ Species * Sepal.Width + Petal.Width + ID + Sepal.Width:ID, data = iris,
rscaleCont = c(Sepal.Width = 0.123, Petal.Width = 0.321),
whichRandom = c("ID","Sepal.Width:ID"))
#> Warning in rscale[rscaleTypes == "continuous"] <- rscaleCont: number of items to
#> replace is not a multiple of replacement length
#> Warning in doNwaySampling(method, y, X, rscale, iterations, gMap, incCont, :
#> Some NAs were removed from sampling results: 10000 in total.
clean_parameters(m1)
#> Warning in rscale[rscaleTypes == "continuous"] <- rscaleCont: number of items to
#> replace is not a multiple of replacement length
#> Parameter Effects Component
#> 1 Species-setosa fixed conditional
#> 2 Species-versicolor fixed conditional
#> 3 Species-virginica fixed conditional
#> 4 Sepal.Width fixed conditional
#> 5 Petal.Width fixed conditional
#> 6 ID-1 fixed conditional
#> 7 ID-2 fixed conditional
#> 8 ID-3 fixed conditional
#> 9 ID-4 fixed conditional
#> 10 ID-5 fixed conditional
#> 11 ID-6 fixed conditional
#> 12 ID-7 fixed conditional
#> 13 ID-8 fixed conditional
#> 14 ID-9 fixed conditional
#> 15 ID-10 fixed conditional
#> 16 ID-11 fixed conditional
#> 17 ID-12 fixed conditional
#> 18 ID-13 fixed conditional
#> 19 ID-14 fixed conditional
#> 20 ID-15 fixed conditional
#> 21 ID-16 fixed conditional
#> 22 ID-17 fixed conditional
#> 23 ID-18 fixed conditional
#> 24 ID-19 fixed conditional
#> 25 ID-20 fixed conditional
#> 26 ID-21 fixed conditional
#> 27 ID-22 fixed conditional
#> 28 ID-23 fixed conditional
#> 29 ID-24 fixed conditional
#> 30 ID-25 fixed conditional
#> 31 ID-26 fixed conditional
#> 32 ID-27 fixed conditional
#> 33 ID-28 fixed conditional
#> 34 ID-29 fixed conditional
#> 35 ID-30 fixed conditional
#> 36 Species:Sepal.Width-setosa.&.Sepal.Width fixed conditional
#> 37 Species:Sepal.Width-versicolor.&.Sepal.Width fixed conditional
#> 38 Species:Sepal.Width-virginica.&.Sepal.Width fixed conditional
#> 39 Sepal.Width:ID-1 fixed conditional
#> 40 Sepal.Width:ID-2 fixed conditional
#> 41 Sepal.Width:ID-3 fixed conditional
#> 42 Sepal.Width:ID-4 fixed conditional
#> 43 Sepal.Width:ID-5 fixed conditional
#> 44 Sepal.Width:ID-6 fixed conditional
#> 45 Sepal.Width:ID-7 fixed conditional
#> 46 Sepal.Width:ID-8 fixed conditional
#> 47 Sepal.Width:ID-9 fixed conditional
#> 48 Sepal.Width:ID-10 fixed conditional
#> 49 Sepal.Width:ID-11 fixed conditional
#> 50 Sepal.Width:ID-12 fixed conditional
#> 51 Sepal.Width:ID-13 fixed conditional
#> 52 Sepal.Width:ID-14 fixed conditional
#> 53 Sepal.Width:ID-15 fixed conditional
#> 54 Sepal.Width:ID-16 fixed conditional
#> 55 Sepal.Width:ID-17 fixed conditional
#> 56 Sepal.Width:ID-18 fixed conditional
#> 57 Sepal.Width:ID-19 fixed conditional
#> 58 Sepal.Width:ID-20 fixed conditional
#> 59 Sepal.Width:ID-21 fixed conditional
#> 60 Sepal.Width:ID-22 fixed conditional
#> 61 Sepal.Width:ID-23 fixed conditional
#> 62 Sepal.Width:ID-24 fixed conditional
#> 63 Sepal.Width:ID-25 fixed conditional
#> 64 Sepal.Width:ID-26 fixed conditional
#> 65 Sepal.Width:ID-27 fixed conditional
#> 66 Sepal.Width:ID-28 fixed conditional
#> 67 Sepal.Width:ID-29 fixed conditional
#> 68 Sepal.Width:ID-30 fixed conditional
#> 69 ID-1 random conditional
#> 70 ID-2 random conditional
#> 71 ID-3 random conditional
#> 72 ID-4 random conditional
#> 73 ID-5 random conditional
#> 74 ID-6 random conditional
#> 75 ID-7 random conditional
#> 76 ID-8 random conditional
#> 77 ID-9 random conditional
#> 78 ID-10 random conditional
#> 79 ID-11 random conditional
#> 80 ID-12 random conditional
#> 81 ID-13 random conditional
#> 82 ID-14 random conditional
#> 83 ID-15 random conditional
#> 84 ID-16 random conditional
#> 85 ID-17 random conditional
#> 86 ID-18 random conditional
#> 87 ID-19 random conditional
#> 88 ID-20 random conditional
#> 89 ID-21 random conditional
#> 90 ID-22 random conditional
#> 91 ID-23 random conditional
#> 92 ID-24 random conditional
#> 93 ID-25 random conditional
#> 94 ID-26 random conditional
#> 95 ID-27 random conditional
#> 96 ID-28 random conditional
#> 97 ID-29 random conditional
#> 98 ID-30 random conditional
#> 99 mu fixed extra
#> 100 Sepal.Width-Sepal.Width fixed extra
#> 101 Petal.Width-Petal.Width fixed extra
#> 102 sig2 fixed extra
#> 103 g_Species fixed extra
#> 104 g_ID fixed extra
#> 105 g_continuous fixed extra
#> Cleaned_Parameter
#> 1 Species-setosa
#> 2 Species-versicolor
#> 3 Species-virginica
#> 4 Sepal.Width
#> 5 Petal.Width
#> 6 ID-1
#> 7 ID-2
#> 8 ID-3
#> 9 ID-4
#> 10 ID-5
#> 11 ID-6
#> 12 ID-7
#> 13 ID-8
#> 14 ID-9
#> 15 ID-10
#> 16 ID-11
#> 17 ID-12
#> 18 ID-13
#> 19 ID-14
#> 20 ID-15
#> 21 ID-16
#> 22 ID-17
#> 23 ID-18
#> 24 ID-19
#> 25 ID-20
#> 26 ID-21
#> 27 ID-22
#> 28 ID-23
#> 29 ID-24
#> 30 ID-25
#> 31 ID-26
#> 32 ID-27
#> 33 ID-28
#> 34 ID-29
#> 35 ID-30
#> 36 Species:Sepal.Width-setosa.&.Sepal.Width
#> 37 Species:Sepal.Width-versicolor.&.Sepal.Width
#> 38 Species:Sepal.Width-virginica.&.Sepal.Width
#> 39 Sepal.Width:ID-1
#> 40 Sepal.Width:ID-2
#> 41 Sepal.Width:ID-3
#> 42 Sepal.Width:ID-4
#> 43 Sepal.Width:ID-5
#> 44 Sepal.Width:ID-6
#> 45 Sepal.Width:ID-7
#> 46 Sepal.Width:ID-8
#> 47 Sepal.Width:ID-9
#> 48 Sepal.Width:ID-10
#> 49 Sepal.Width:ID-11
#> 50 Sepal.Width:ID-12
#> 51 Sepal.Width:ID-13
#> 52 Sepal.Width:ID-14
#> 53 Sepal.Width:ID-15
#> 54 Sepal.Width:ID-16
#> 55 Sepal.Width:ID-17
#> 56 Sepal.Width:ID-18
#> 57 Sepal.Width:ID-19
#> 58 Sepal.Width:ID-20
#> 59 Sepal.Width:ID-21
#> 60 Sepal.Width:ID-22
#> 61 Sepal.Width:ID-23
#> 62 Sepal.Width:ID-24
#> 63 Sepal.Width:ID-25
#> 64 Sepal.Width:ID-26
#> 65 Sepal.Width:ID-27
#> 66 Sepal.Width:ID-28
#> 67 Sepal.Width:ID-29
#> 68 Sepal.Width:ID-30
#> 69 ID-1
#> 70 ID-2
#> 71 ID-3
#> 72 ID-4
#> 73 ID-5
#> 74 ID-6
#> 75 ID-7
#> 76 ID-8
#> 77 ID-9
#> 78 ID-10
#> 79 ID-11
#> 80 ID-12
#> 81 ID-13
#> 82 ID-14
#> 83 ID-15
#> 84 ID-16
#> 85 ID-17
#> 86 ID-18
#> 87 ID-19
#> 88 ID-20
#> 89 ID-21
#> 90 ID-22
#> 91 ID-23
#> 92 ID-24
#> 93 ID-25
#> 94 ID-26
#> 95 ID-27
#> 96 ID-28
#> 97 ID-29
#> 98 ID-30
#> 99 mu
#> 100 Sepal.Width-Sepal.Width
#> 101 Petal.Width-Petal.Width
#> 102 sig2
#> 103 g_Species
#> 104 g_ID
#> 105 g_continuous
Created on 2020-09-17 by the reprex package (v0.3.0)
The issue in the above examples is:
> Warning in rscale[rscaleTypes == "continuous"] <- rscaleCont: number of items to
> replace is not a multiple of replacement length
Any possibility that this can be resolved before the next round (https://github.com/easystats/easystats/issues/76)?
I can then also do a next round of release for ggstatsverse
packages then ๐
Yes, I think so. Next update for insight is still some weeks ahead.
Oh no, I forgot that this also needs to be resolved for the next tidyBF
release ๐
I looked into this some days ago, but it's something that should be addressed with some concentration, so I skipped this issue for this release and will get back to it when I find more time.
Yeah, no worries. I have a workaround solution using bayestestR::describe_prior
, so all good for now.
Reminder for ourselves to also add then Prior_scale
info for meta_random
and meta_fixed
objects in model_parameters
output once this is resolved:
library(metaBMA)
#> Loading required package: Rcpp
data(towels)
set.seed(123)
mr <-
meta_random(
logOR,
SE,
study,
data = towels
)
#> Warning: There were 4 divergent transitions after warmup. See
#> http://mc-stan.org/misc/warnings.html#divergent-transitions-after-warmup
#> to find out why this is a problem and how to eliminate them.
#> Warning: Examine the pairs() plot to diagnose sampling problems
mr$jzs
#> $rscale_contin
#> [1] 0.5
#>
#> $rscale_discrete
#> [1] 0.7071068
#>
#> $centering
#> [1] TRUE
parameters::model_parameters(mr)
#> Parameter | Coefficient | SE | CI | BF | Rhat | ESS
#> -----------------------------------------------------------------------
#> Overall | 0.18 | 0.10 | [-0.02, 0.38] | 2.00 | 1.000 | 3974.80
#> tau | 0.14 | 0.10 | [ 0.02, 0.33] | | 1.000 | 3145.40
Created on 2020-10-26 by the reprex package (v0.3.0.9001)
I think there's not much to do on insight side:
library(metaBMA)
#> Loading required package: Rcpp
data(towels)
set.seed(123)
mr <-
meta_random(
logOR,
SE,
study,
data = towels
)
#> Warning: There were 4 divergent transitions after warmup. Increasing adapt_delta above 0.95 may help. See
#> http://mc-stan.org/misc/warnings.html#divergent-transitions-after-warmup
#> Warning: Examine the pairs() plot to diagnose sampling problems
insight::get_priors(mr)
#> Parameter Distribution Location Scale
#> 1 (Intercept) norm 0 0.30
#> 2 tau invgamma 1 0.15
Created on 2020-10-26 by the reprex package (v0.3.0)
However, I'm not sure where the formatting of priors is taking place, bayestestR?
Why are the values from insight::get_priors
here different than that contained in the object itself?
mr$jzs
#> $rscale_contin
#> [1] 0.5
#>
#> $rscale_discrete
#> [1] 0.7071068
#>
#> $centering
#> [1] TRUE
Not sure what jzs is, but get_priors()
is consistent with the print()
-method:
library(metaBMA)
#> Loading required package: Rcpp
data(towels)
set.seed(123)
mr <-
meta_random(
logOR,
SE,
study,
data = towels
)
insight::get_priors(mr)
#> Parameter Distribution Location Scale
#> 1 (Intercept) norm 0 0.30
#> 2 tau invgamma 1 0.15
mr
#> ### Bayesian Random-Effects Meta-Analysis ###
#> Prior on d: 'norm' (mean=0, sd=0.3) with support on the interval [-Inf,Inf].
#> Prior on tau: 'invgamma' (shape=1, scale=0.15) with support on the interval [0,Inf].
#>
#> # Bayes factors:
#> (denominator)
#> (numerator) random_H0 random_H1
#> random_H0 1 0.501
#> random_H1 2 1.000
#>
#> # Posterior summary statistics of random-effects model:
#> mean sd 2.5% 50% 97.5% hpd95_lower hpd95_upper n_eff Rhat
#> d 0.182 0.104 -0.039 0.188 0.367 -0.021 0.381 3974.8 1
#> tau 0.136 0.099 0.033 0.108 0.404 0.020 0.330 3145.4 1
Created on 2020-10-26 by the reprex package (v0.3.0)
They are the priors used in JZS method for calculating Bayes Factors (e.g., https://link.springer.com/article/10.3758/PBR.16.2.225).
BayesFactor
package also uses this method.
But I also see your point. I am not sure what the best way to report priors here would be.
@mattansb Do you have thoughts on this?
If I had to guess, insight::get_priors(mr)
gives the raw priors (on the response scale) and mr$jzs
gives the standardized priors.
Maybe someone can file a PR?
Any chance this can be fixed in the next release?
The "next release" will be within the next ~24 hours, so I'd say, no. ๐ฌ
Any possibility we can get this fixed before the next release cycle?
I can't even remember the exact issue we had here... ^^
The issue was to match a prior to each parameter ๐งถ
library(parameters)
library(BayesFactor)
iris$ID <- factor(rep(1:30, each = 5))
m1 <- lmBF(Sepal.Length ~ Species * Sepal.Width + Petal.Width + ID + Sepal.Width:ID, data = iris,
rscaleCont = c(Sepal.Width = 0.123, Petal.Width = 0.321),
whichRandom = c("ID","Sepal.Width:ID"))
model_parameters(m1)
#> # Extra Parameters
#>
#> Parameter | pd | Prior
#> -----------------------------------------------------
#> mu | 98.95% |
#> Sepal.Width-Sepal.Width | 99.12% | Cauchy (0 +- 0.12)
#> Petal.Width-Petal.Width | 99.08% | Cauchy (0 +- 0.32)
#> sig2 | 100% |
#> g_Species | 100% |
#> g_ID | 100% |
#> g_continuous | 100% |
#>
#> # Fixed Effects
#>
#> Parameter | pd | Prior
#> --------------------------------------------------------------------------
#> Species-setosa | 98.90% | Cauchy (0 +- 0.50)
#> Species-versicolor | 98.98% | Cauchy (0 +- 0.50)
#> Species-virginica | 98.98% | Cauchy (0 +- 0.50)
#> ID-1 | 99.00% | Cauchy (0 +- 1.00)
#> ID-1 | 99.00% | Cauchy (0 +- 1.00)
#> ID-2 | 98.90% | Cauchy (0 +- 1.00)
#> ID-2 | 98.90% | Cauchy (0 +- 1.00)
#> ID-3 | 98.98% | Cauchy (0 +- 1.00)
#> ID-3 | 98.98% | Cauchy (0 +- 1.00)
#> ID-4 | 98.98% | Cauchy (0 +- 1.00)
#> ID-4 | 98.98% | Cauchy (0 +- 1.00)
#> ID-5 | 98.92% | Cauchy (0 +- 1.00)
#> ID-5 | 98.92% | Cauchy (0 +- 1.00)
#> ID-6 | 99.08% | Cauchy (0 +- 1.00)
#> ID-6 | 99.08% | Cauchy (0 +- 1.00)
#> ID-7 | 98.98% | Cauchy (0 +- 1.00)
#> ID-7 | 98.98% | Cauchy (0 +- 1.00)
#> ID-8 | 98.98% | Cauchy (0 +- 1.00)
#> ID-8 | 98.98% | Cauchy (0 +- 1.00)
#> ID-9 | 98.95% | Cauchy (0 +- 1.00)
#> ID-9 | 98.95% | Cauchy (0 +- 1.00)
#> ID-10 | 98.98% | Cauchy (0 +- 1.00)
#> ID-10 | 98.98% | Cauchy (0 +- 1.00)
#> ID-11 | 98.98% | Cauchy (0 +- 1.00)
#> ID-11 | 98.98% | Cauchy (0 +- 1.00)
#> ID-12 | 98.90% | Cauchy (0 +- 1.00)
#> ID-12 | 98.90% | Cauchy (0 +- 1.00)
#> ID-13 | 99.08% | Cauchy (0 +- 1.00)
#> ID-13 | 99.08% | Cauchy (0 +- 1.00)
#> ID-14 | 99.10% | Cauchy (0 +- 1.00)
#> ID-14 | 99.10% | Cauchy (0 +- 1.00)
#> ID-15 | 99.02% | Cauchy (0 +- 1.00)
#> ID-15 | 99.02% | Cauchy (0 +- 1.00)
#> ID-16 | 99.02% | Cauchy (0 +- 1.00)
#> ID-16 | 99.02% | Cauchy (0 +- 1.00)
#> ID-17 | 99.00% | Cauchy (0 +- 1.00)
#> ID-17 | 99.00% | Cauchy (0 +- 1.00)
#> ID-18 | 99.17% | Cauchy (0 +- 1.00)
#> ID-18 | 99.17% | Cauchy (0 +- 1.00)
#> ID-19 | 99.02% | Cauchy (0 +- 1.00)
#> ID-19 | 99.02% | Cauchy (0 +- 1.00)
#> ID-20 | 98.90% | Cauchy (0 +- 1.00)
#> ID-20 | 98.90% | Cauchy (0 +- 1.00)
#> ID-21 | 98.90% | Cauchy (0 +- 1.00)
#> ID-21 | 98.90% | Cauchy (0 +- 1.00)
#> ID-22 | 99.02% | Cauchy (0 +- 1.00)
#> ID-22 | 99.02% | Cauchy (0 +- 1.00)
#> ID-23 | 99.02% | Cauchy (0 +- 1.00)
#> ID-23 | 99.02% | Cauchy (0 +- 1.00)
#> ID-24 | 98.95% | Cauchy (0 +- 1.00)
#> ID-24 | 98.95% | Cauchy (0 +- 1.00)
#> ID-25 | 99.00% | Cauchy (0 +- 1.00)
#> ID-25 | 99.00% | Cauchy (0 +- 1.00)
#> ID-26 | 99.12% | Cauchy (0 +- 1.00)
#> ID-26 | 99.12% | Cauchy (0 +- 1.00)
#> ID-27 | 98.95% | Cauchy (0 +- 1.00)
#> ID-27 | 98.95% | Cauchy (0 +- 1.00)
#> ID-28 | 98.92% | Cauchy (0 +- 1.00)
#> ID-28 | 98.92% | Cauchy (0 +- 1.00)
#> ID-29 | 98.95% | Cauchy (0 +- 1.00)
#> ID-29 | 98.95% | Cauchy (0 +- 1.00)
#> ID-30 | 99.02% | Cauchy (0 +- 1.00)
#> ID-30 | 99.02% | Cauchy (0 +- 1.00)
#> Species:Sepal.Width-setosa.&.Sepal.Width | 98.95% | Cauchy (0 +- 0.12)
#> Species:Sepal.Width-versicolor.&.Sepal.Width | 98.92% | Cauchy (0 +- 0.12)
#> Species:Sepal.Width-virginica.&.Sepal.Width | 99.05% | Cauchy (0 +- 0.12)
#> Sepal.Width:ID-1 | 99.08% | Cauchy (0 +- 1.00)
#> Sepal.Width:ID-2 | 98.92% | Cauchy (0 +- 1.00)
#> Sepal.Width:ID-3 | 98.92% | Cauchy (0 +- 1.00)
#> Sepal.Width:ID-4 | 99.10% | Cauchy (0 +- 1.00)
#> Sepal.Width:ID-5 | 98.90% | Cauchy (0 +- 1.00)
#> Sepal.Width:ID-6 | 99.10% | Cauchy (0 +- 1.00)
#> Sepal.Width:ID-7 | 98.98% | Cauchy (0 +- 1.00)
#> Sepal.Width:ID-8 | 99.05% | Cauchy (0 +- 1.00)
#> Sepal.Width:ID-9 | 99.08% | Cauchy (0 +- 1.00)
#> Sepal.Width:ID-10 | 98.92% | Cauchy (0 +- 1.00)
#> Sepal.Width:ID-11 | 98.92% | Cauchy (0 +- 1.00)
#> Sepal.Width:ID-12 | 99.05% | Cauchy (0 +- 1.00)
#> Sepal.Width:ID-13 | 98.98% | Cauchy (0 +- 1.00)
#> Sepal.Width:ID-14 | 98.92% | Cauchy (0 +- 1.00)
#> Sepal.Width:ID-15 | 99.17% | Cauchy (0 +- 1.00)
#> Sepal.Width:ID-16 | 98.98% | Cauchy (0 +- 1.00)
#> Sepal.Width:ID-17 | 99.10% | Cauchy (0 +- 1.00)
#> Sepal.Width:ID-18 | 99.17% | Cauchy (0 +- 1.00)
#> Sepal.Width:ID-19 | 98.98% | Cauchy (0 +- 1.00)
#> Sepal.Width:ID-20 | 99.00% | Cauchy (0 +- 1.00)
#> Sepal.Width:ID-21 | 99.00% | Cauchy (0 +- 1.00)
#> Sepal.Width:ID-22 | 98.98% | Cauchy (0 +- 1.00)
#> Sepal.Width:ID-23 | 98.98% | Cauchy (0 +- 1.00)
#> Sepal.Width:ID-24 | 99.20% | Cauchy (0 +- 1.00)
#> Sepal.Width:ID-25 | 99.17% | Cauchy (0 +- 1.00)
#> Sepal.Width:ID-26 | 98.95% | Cauchy (0 +- 1.00)
#> Sepal.Width:ID-27 | 98.95% | Cauchy (0 +- 1.00)
#> Sepal.Width:ID-28 | 98.92% | Cauchy (0 +- 1.00)
#> Sepal.Width:ID-29 | 98.98% | Cauchy (0 +- 1.00)
#> Sepal.Width:ID-30 | 99.10% | Cauchy (0 +- 1.00)
#> Petal.Width | | Cauchy (0 +- 0.32)
#> Sepal.Width | | Cauchy (0 +- 0.12)
#>
#> # Random Effects
#>
#> Parameter | pd | Prior
#> ------------------------------------
#> ID-1 | 99.00% | Cauchy (0 +- 1)
#> ID-1 | 99.00% | Cauchy (0 +- 1)
#> ID-2 | 98.90% | Cauchy (0 +- 1)
#> ID-2 | 98.90% | Cauchy (0 +- 1)
#> ID-3 | 98.98% | Cauchy (0 +- 1)
#> ID-3 | 98.98% | Cauchy (0 +- 1)
#> ID-4 | 98.98% | Cauchy (0 +- 1)
#> ID-4 | 98.98% | Cauchy (0 +- 1)
#> ID-5 | 98.92% | Cauchy (0 +- 1)
#> ID-5 | 98.92% | Cauchy (0 +- 1)
#> ID-6 | 99.08% | Cauchy (0 +- 1)
#> ID-6 | 99.08% | Cauchy (0 +- 1)
#> ID-7 | 98.98% | Cauchy (0 +- 1)
#> ID-7 | 98.98% | Cauchy (0 +- 1)
#> ID-8 | 98.98% | Cauchy (0 +- 1)
#> ID-8 | 98.98% | Cauchy (0 +- 1)
#> ID-9 | 98.95% | Cauchy (0 +- 1)
#> ID-9 | 98.95% | Cauchy (0 +- 1)
#> ID-10 | 98.98% | Cauchy (0 +- 1)
#> ID-10 | 98.98% | Cauchy (0 +- 1)
#> ID-11 | 98.98% | Cauchy (0 +- 1)
#> ID-11 | 98.98% | Cauchy (0 +- 1)
#> ID-12 | 98.90% | Cauchy (0 +- 1)
#> ID-12 | 98.90% | Cauchy (0 +- 1)
#> ID-13 | 99.08% | Cauchy (0 +- 1)
#> ID-13 | 99.08% | Cauchy (0 +- 1)
#> ID-14 | 99.10% | Cauchy (0 +- 1)
#> ID-14 | 99.10% | Cauchy (0 +- 1)
#> ID-15 | 99.02% | Cauchy (0 +- 1)
#> ID-15 | 99.02% | Cauchy (0 +- 1)
#> ID-16 | 99.02% | Cauchy (0 +- 1)
#> ID-16 | 99.02% | Cauchy (0 +- 1)
#> ID-17 | 99.00% | Cauchy (0 +- 1)
#> ID-17 | 99.00% | Cauchy (0 +- 1)
#> ID-18 | 99.17% | Cauchy (0 +- 1)
#> ID-18 | 99.17% | Cauchy (0 +- 1)
#> ID-19 | 99.02% | Cauchy (0 +- 1)
#> ID-19 | 99.02% | Cauchy (0 +- 1)
#> ID-20 | 98.90% | Cauchy (0 +- 1)
#> ID-20 | 98.90% | Cauchy (0 +- 1)
#> ID-21 | 98.90% | Cauchy (0 +- 1)
#> ID-21 | 98.90% | Cauchy (0 +- 1)
#> ID-22 | 99.02% | Cauchy (0 +- 1)
#> ID-22 | 99.02% | Cauchy (0 +- 1)
#> ID-23 | 99.02% | Cauchy (0 +- 1)
#> ID-23 | 99.02% | Cauchy (0 +- 1)
#> ID-24 | 98.95% | Cauchy (0 +- 1)
#> ID-24 | 98.95% | Cauchy (0 +- 1)
#> ID-25 | 99.00% | Cauchy (0 +- 1)
#> ID-25 | 99.00% | Cauchy (0 +- 1)
#> ID-26 | 99.12% | Cauchy (0 +- 1)
#> ID-26 | 99.12% | Cauchy (0 +- 1)
#> ID-27 | 98.95% | Cauchy (0 +- 1)
#> ID-27 | 98.95% | Cauchy (0 +- 1)
#> ID-28 | 98.92% | Cauchy (0 +- 1)
#> ID-28 | 98.92% | Cauchy (0 +- 1)
#> ID-29 | 98.95% | Cauchy (0 +- 1)
#> ID-29 | 98.95% | Cauchy (0 +- 1)
#> ID-30 | 99.02% | Cauchy (0 +- 1)
#> ID-30 | 99.02% | Cauchy (0 +- 1)
data(puzzles)
result = anovaBF(RT ~ shape*color + ID, data = puzzles, whichRandom = "ID",
whichModels = 'top', progress=FALSE)
model_parameters(result)
#> Multiple `BFBayesFactor` models detected - posteriors are extracted from the first numerator model.
#> See help("get_parameters", package = "insight").
#> Multiple `BFBayesFactor` models detected - posteriors are extracted from the first numerator model.
#> See help("get_parameters", package = "insight").
#> # Extra Parameters
#>
#> Parameter | Median | 89% CI | pd | % in ROPE | BF
#> --------------------------------------------------------------
#> mu | 45.00 | [43.91, 46.10] | 100% | 0% | 2.75
#> sig2 | 1.76 | [ 1.12, 2.45] | 100% | 0% | 0.265
#> g_shape | 0.35 | [ 0.01, 2.23] | 100% | 44.73% | 2.75
#> g_color | 0.36 | [ 0.02, 2.32] | 100% | 43.67% | 0.253
#> g_ID | 2.64 | [ 0.76, 5.23] | 100% | 0% | 0.265
#>
#> # Fixed Effects
#>
#> Parameter | Median | 89% CI | pd | % in ROPE | Prior | BF
#> -------------------------------------------------------------------------------------------------
#> shape [round] | 0.43 | [ 0.14, 0.74] | 99.00% | 12.69% | Cauchy (0 +- 0.50) | 0.253
#> shape [square] | -0.43 | [-0.74, -0.14] | 99.00% | 12.69% | Cauchy (0 +- 0.50) | 0.265
#> color | -0.43 | [-0.73, -0.14] | 98.92% | 12.66% | Cauchy (0 +- 0.50) | 2.75
#> color [monochromatic] | 0.43 | [ 0.14, 0.73] | 98.92% | 12.66% | Cauchy (0 +- 0.50) | 0.253
#>
#> # Random Effects
#>
#> Parameter | Median | 89% CI | pd | % in ROPE | Prior | BF
#> ----------------------------------------------------------------------------------
#> ID [1] | 2.48 | [ 1.11, 3.96] | 99.62% | 0% | Cauchy (0 +- 1) | 0.265
#> ID [2] | 0.48 | [-0.87, 1.99] | 70.33% | 21.68% | Cauchy (0 +- 1) | 2.75
#> ID [3] | 0.92 | [-0.42, 2.43] | 85.50% | 15.45% | Cauchy (0 +- 1) | 0.253
#> ID [4] | 0.46 | [-0.94, 1.95] | 70.03% | 22.04% | Cauchy (0 +- 1) | 0.265
#> ID [5] | 3.19 | [ 1.75, 4.64] | 99.98% | 0% | Cauchy (0 +- 1) | 2.75
#> ID [6] | 0.47 | [-0.93, 1.86] | 69.88% | 22.52% | Cauchy (0 +- 1) | 0.253
#> ID [7] | -3.14 | [-4.60, -1.70] | 99.98% | 0% | Cauchy (0 +- 1) | 0.265
#> ID [8] | -0.18 | [-1.70, 1.19] | 58.27% | 25.67% | Cauchy (0 +- 1) | 2.75
#> ID [9] | -2.47 | [-3.98, -1.07] | 99.67% | 0% | Cauchy (0 +- 1) | 0.253
#> ID [10] | 0.69 | [-0.82, 2.03] | 77.03% | 19.77% | Cauchy (0 +- 1) | 0.265
#> ID [11] | 0.67 | [-0.73, 2.13] | 77.62% | 18.81% | Cauchy (0 +- 1) | 2.75
#> ID [12] | -3.35 | [-4.95, -2.03] | 99.98% | 0% | Cauchy (0 +- 1) | 0.253
Created on 2021-01-03 by the reprex package (v0.3.0)
In the first example,
#> mu | 98.95% |
#> Sepal.Width-Sepal.Width | 99.12% | Cauchy (0 +- 0.12)
#> Petal.Width-Petal.Width | 99.08% | Cauchy (0 +- 0.32)
Are fixed effects, and all the ID and Sepal.Width:ID parameters are random effects (some of which are doubled an also appear in the random effects part.
@mattansb Thinking of this further, another thing I notice is that
Prior_Scale
values areNA
foranovaBF
outputs.Should we be including
rscaleFixed
andrscaleRandom
argument values here?