我正在尝试建立一个既包含对象又包含主题因素的数据模型。尝试使用{
role: "admin"
}进行功效分析和样本量估算时遇到一些问题。
以下是一些模拟数据和模型来说明我的问题：
simr我想知道我是否没有正确指定## List of required packages
Pkgs <- c("tidyverse","magrittr", "lme4", "lmerTest", "MCMCglmm", "simr")
# Load packages
lapply(Pkgs, require, c = T)
## Build a simulated data set
Sub_Group <- data.frame(Subject = LETTERS[1:20],
Group = c(rep("Y", 10), rep("Z", 10)))
Cat_Item <- data.frame(expand.grid(Category = c(rep("U", 25), rep("V", 25)),
Item = factor(rep(c(1:25),2))))
Cond <- data.frame(Condition = letters[1:5])
Cond_Cat_Item <- merge(Cond, Cat_Item, by = NULL)
Data <- merge(Sub_Group, Cond_Cat_Item, by = NULL) %>%
mutate(Y = round(rtnorm(nrow(.), 4.5, 2, 1, 9), digits = 0))
## Run initial glmer without any specified starting values
summary(Mod1 <- glmer(Y ~ Category + Condition + Group +
(1 + Group|Subject),
data = Data,
family = gaussian(link = "log")))
plot(powerCurve(extend(Mod1, along = "Subject", n = 50), along = "Subject", nsim = 10))
# Errors suggesting to provide starting values
lastResult()$errors
# "cannot find valid starting values: please specify some"
## Second glmer with starting values
summary(Mod2 <- glmer(Y ~ Category + Condition + Group +
(1 + Group|Subject),
data = Data,
mustart = Data$Y,
family = gaussian(link = "log")))
plot(powerCurve(extend(Mod2, along = "Subject", n = 50), along = "Subject", nsim = 10))
## Errors regarding variable lengths
lastResult()$errors
# "variable lengths differ (found for '(mustart)')"
值（即使模型运行正常），或者是否有问题或我应该为mustart做些什么。 / p>