Conditional_effects.brmsfit

Interfaces brms
1

Dear all,

I am performing a meta-analysis in brms. My data is correlated, so I’m using a variance-covariance matrix. However, when I use the variance-covariance matrix and the conditional_effects
function, the following error appears: Error in get_dpar(object, dpar = dp) : !is.null(x) is not TRUE.
However, when I fit the model and I don’t use the variance-covariance matrix, I can plot the figures using the conditional_effects function.

2

Hi and welcome! Would it be possible for you to post the model code you are running and either a snippet of the data or simulated data. Also your version of R and brms would be helpful.

thanks!

3

Hi, @Ara_Winter thanks for your answer.

  • R version 3.6.1
  • brms Version: 2.11.1

The dataset:
dataset.csv (660 Bytes)

The coding:

# covariance matrix of outcomes
   cor_matrix <- function(x, r, v = rep(1, length(x)), na.rm = FALSE) {
   mat <- diag(v)
   se <- sqrt(v)
   se[is.na(se)] <- 0
   if (length(x) > 1L) {
      for (i in 2:nrow(mat)) {
         for (j in 1:(i-1)) {
            if (x[i] == x[j]) {
               mat[i, j] <- mat[j, i] <- r * se[i] * se[j] 
            }
         }
      }
   }
   dimnames(mat) <- list(1:nrow(mat), 1:ncol(mat))
   if (na.rm) {
      keep <- !is.na(diag(mat))
      mat <- mat[keep, keep]
   }
   mat
}

effect_coding <- function(x, levels) {
   # convenience function for effect (i.e. sum) coding
   x <- factor(x, levels = levels)
   contrasts(x) <- contr.sum(length(levels))
   colnames(contrasts(x)) <- levels[-length(levels)]
   x
}


prior <-c(prior(normal(0, 100), class = Intercept),
          prior(normal(0, 10), class = b),
          prior(cauchy(0, 10), class = sd))
                                                 

v.m1 <- cor_matrix(dat$study_ID, r = 0.9, v = dat$vi)
cor_obs <- cor_matrix(dat$study_ID, r = 0.9)

m1 <- brm(yi ~ 1 + log_time + level + exer +
             (1|study_ID), 
             family = gaussian(), 
             prior = prior, autocor = cor_fixed(v.m1),  
             cores = parallel::detectCores(),
             control = list(adapt_delta = 0.99), data = dat)

    conditions1 <- data.frame(level = 1)
    plot(conditional_effects(m1, effects = "log_time:exer", 
                         conditions = conditions1))
1 Like
4

Thanks! I am hoping someone more proficient in brms will be able to answer. I’ll see if I can tag some people in.

1 Like
5

I know the dev @paul.buerkner posts here but I wasn’t sure if there is a list of brms users who can also help answer this question.

6

Sorry, I have overlooked this post. Does it still fail in the latest github version of brms?

Note that, in the github version, I recommend specifying the autocor term directly in formula via

... ~ ... + fcor(v.m1)

instead of in autocor. Please also note that the residual SD sigma will be estimated by default in this case. To turns it off, specify

formula = bf(
  yi ~ 1 + log_time + level + exer + (1|study_ID) + fcor(v.m1),
  sigma = 1
)
1 Like
7

Dear @paul.buerkner, thanks for your response. Yes, it still fails in the latest GitHub version of brms. I followed your suggestion, using the formula function, as below:

formula = bf(
   yi ~ 1 + log_time + level + exer + (1|study_ID) + fcor(v.m1), data = dat,
   sigma = 1

)
and the following error message appeared:
Error: Expecting a single value when fixing parameter 'data.pred_ID'.

8

Ok. Can you please provide a minimal reproducible example. I don’t see definition of dat in the provided code.

9

With the following code, it works for me:

# covariance matrix of outcomes
cor_matrix <- function(x, r, v = rep(1, length(x)), na.rm = FALSE) {
  mat <- diag(v)
  se <- sqrt(v)
  se[is.na(se)] <- 0
  if (length(x) > 1L) {
    for (i in 2:nrow(mat)) {
      for (j in 1:(i-1)) {
        if (x[i] == x[j]) {
          mat[i, j] <- mat[j, i] <- r * se[i] * se[j] 
        }
      }
    }
  }
  dimnames(mat) <- list(1:nrow(mat), 1:ncol(mat))
  if (na.rm) {
    keep <- !is.na(diag(mat))
    mat <- mat[keep, keep]
  }
  mat
}

effect_coding <- function(x, levels) {
  # convenience function for effect (i.e. sum) coding
  x <- factor(x, levels = levels)
  contrasts(x) <- contr.sum(length(levels))
  colnames(contrasts(x)) <- levels[-length(levels)]
  x
}

dat <- data.frame(
  yi = rnorm(20),
  level = factor(rep(0:1, each = 10)),
  log_time = rnorm(20),
  exer = rnorm(20),
  study_ID = 1:20,
  vi = 0.25
)

prior <-c(prior(normal(0, 100), class = Intercept),
          prior(normal(0, 10), class = b),
          prior(cauchy(0, 10), class = sd))


v.m1 <- cor_matrix(dat$study_ID, r = 0.9, v = dat$vi)
cor_obs <- cor_matrix(dat$study_ID, r = 0.9)

# brms 2.11.6
m1 <- brm(bf(
            yi ~ 1 + log_time + level + exer + (1|study_ID) + fcor(v.m1),
            sigma = 1
          ),
          family = gaussian(), 
          prior = prior, chains = 1,
          data2 = list(v.m1 = v.m1),
          control = list(adapt_delta = 0.99), data = dat)

conditions1 <- data.frame(level = 1)
plot(conditional_effects(m1, effects = "log_time:exer", 
                         conditions = conditions1))
2 Likes
10

Now I used the model you provided and replaced it with my data. The model worked perfectly and the figures were generated by conditional_effects.
Thank you very much @paul.buerkner for your help.

1 Like