能否将"mixed"对象转换为"merModLmerTest"对象以计算GLMM的$R^2$？

Great question! The mixed objects from the afex package (the most common source of this object type) don’t require forced conversion—they actually wrap a merModLmerTest object under the hood that you can access directly. Here’s how to handle this:

• 提取底层的merModLmerTest对象
  The mixed() function returns a list-like object where the full fitted model is stored in the full_model slot. You can access it like this:

  # 假设你的mixed对象名为my_mixed_model
  lmer_test_model <- my_mixed_model$full_model
  

  Now you can run fixef(lmer_test_model) and getME(lmer_test_model, c("theta", "beta")) just like you would with a standard merModLmerTest object.

• 更简便的方法：直接用performance包计算$R^2$
  You don’t even need to extract the model manually. The performance package’s r2_nakagawa() function natively supports afex mixed objects, so you can compute the marginal and conditional $R^2$ in one line:

  library(performance)
  r2_nakagawa(my_mixed_model)
  

  This function implements the Nakagawa & Schielzeth (2013) method directly, so it saves you from writing custom code with fixef() and getME().

• 如果确实需要强制转换（一般没必要）
  In rare cases where the slot access doesn’t work (unlikely with standard afex usage), you can try coercing the object with as():

  lmer_test_model <- as(my_mixed_model, "merModLmerTest")
  

  But again, the $full_model approach is the intended and more reliable way to access the underlying model.

内容的提问来源于stack exchange，提问作者KrisAnathema