Stanarm_betareg: Error related to non-finite value supplied by optim

Modeling
1

I tried to run stan_betareg on my model that assesses how MEI index influences the percent cover of kelps. My model is hierarchical with percent cover measurements being nested. Quadrat being nested into Transect being nested into Site. I keep running into this error message and I thought it could be NA values so I tried omitting them but no luck there. I am not sure what could be causing the problem? Note: Because my dataset has zeros and ones, I changed them into 0.00000001 and 0.999999. Thank you!

model <- stan_betareg(Percent_Cover ~ MEI_Index + (1 + MEI_Index | Quadrat / Transect / Site), data = Nino_16)
Error in optim(par = start, fn = loglikfun, gr = gradfun, method = method,  : 
  non-finite value supplied by optim

Here’s my reproducible dataset:

structure(list(Date = structure(c(10L, 10L, 10L, 10L, 10L, 10L, 
10L, 10L, 10L, 10L), .Label = c("4/30/98", "5/14/18", "5/17/00", 
"5/22/00", "5/24/16", "5/25/17", "5/26/17", "5/29/98", "5/3/18", 
"5/6/16", "5/7/16", "6/12/18", "6/13/99", "6/14/18", "6/21/16", 
"6/22/17", "6/23/16", "6/24/17", "6/3/99", "7/11/18", "7/13/18", 
"7/18/16", "7/22/17", "7/23/16", "7/24/17", "7/26/99", "7/28/99", 
"7/3/97", "8/10/98", "8/15/00", "8/16/00", "8/18/16", "8/20/16", 
"8/20/97", "8/6/98"), class = "factor"), Year = c(2016L, 2016L, 
2016L, 2016L, 2016L, 2016L, 2016L, 2016L, 2016L, 2016L), Month = c(5L, 
5L, 5L, 5L, 5L, 5L, 5L, 5L, 5L, 5L), Site = c(1, 1, 1, 1, 1, 
1, 1, 1, 1, 1), Transect = c(1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L), Quadrat = 1:10, Percent_Cover = c(0.3, 0.51, 0.14, 0.03, 
0.04, 0.17, 0.15, 0.21, 0.16, 0.24), MEI_Index = c(1.3, 1.3, 
1.3, 1.3, 1.3, 1.3, 1.3, 1.3, 1.3, 1.3)), row.names = c(NA, 10L
), class = "data.frame")
2

I didn’t think stan_betareg even worked with lme4 constructs, but stan_glmer should work with family = mgcv::betar as long as you are only modeling the mean function.