绘制具有二元预测变量和二元响应变量的混合效应模型

Question

我正在寻找一种令人愉悦且信息丰富的方式来可视化混合模型，其中响应变量和预测变量都是二元的。

m_0 <- glmer(Preselected_0 ~ N_G_altnt_Q_YN + (N_G_altnt_Q_YN | File / Person_anon), family = "binomial", 
                    data = df)

我使用

plot_model

得到的图是这样的：

library(sjPlot)

plot_model(m_0, type = "eff", terms = c("N_G_altnt_Q_YN"), #pred.type = "fe", ci.lvl = .68, line.size = 1.2,
           title = ""
)

我想要得到的情节是这样的：或者，如果二元预测器不可能或不建议这样做，那么其他一些更具视觉吸引力和信息量的可视化 - 任何帮助都会受到赞赏！

数据：

df <- structure(list(N_G_altnt_Q_YN = structure(c(1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 2L, 1L, 2L, 1L, 2L, 2L, 1L, 1L, 1L, 1L, 2L, 
1L, 1L, 1L, 2L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 1L, 1L, 1L, 1L, 1L, 
2L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 1L, 1L, 1L, 1L, 
1L, 2L, 1L, 2L, 1L, 2L, 2L, 1L, 1L, 2L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 2L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 1L, 1L, 1L, 1L, 2L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 1L, 1L, 1L, 
1L, 1L, 2L, 1L, 2L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 2L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 1L, 1L, 1L, 
1L, 2L, 1L, 1L, 1L, 1L, 1L, 2L, 1L, 2L, 2L, 2L, 2L, 1L, 2L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
2L, 1L, 1L, 1L, 1L, 2L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 1L, 2L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 
1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 1L, 1L, 2L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 1L, 1L, 1L, 1L, 2L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 2L), levels = c("0", "1"), class = "factor"), 
    Preselected_0 = structure(c(1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
    1L, 1L, 1L, 2L, 2L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
    1L, 1L, 1L, 2L, 1L, 2L, 2L, 2L, 2L, 2L, 2L, 1L, 1L, 2L, 2L, 
    2L, 2L, 2L, 1L, 1L, 2L, 1L, 1L, 1L, 2L, 1L, 1L, 2L, 2L, 1L, 
    2L, 1L, 1L, 2L, 1L, 1L, 1L, 2L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
    1L, 1L, 2L, 1L, 2L, 2L, 2L, 2L, 2L, 1L, 2L, 1L, 1L, 2L, 2L, 
    2L, 1L, 1L, 2L, 1L, 2L, 1L, 2L, 1L, 2L, 2L, 2L, 1L, 1L, 1L, 
    1L, 1L, 1L, 1L, 2L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 1L, 
    2L, 1L, 2L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 1L, 1L, 1L, 1L, 1L, 
    1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
    1L, 1L, 2L, 1L, 1L, 1L, 1L, 1L, 2L, 1L, 2L, 1L, 1L, 1L, 1L, 
    2L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 1L, 1L, 1L, 1L, 1L, 1L, 
    2L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
    2L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
    1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 1L, 2L, 
    1L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 1L, 2L, 1L, 1L, 1L, 1L, 2L, 
    2L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 1L, 1L, 
    1L, 2L, 2L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 2L, 2L, 1L, 1L, 
    1L, 1L, 1L, 1L, 1L, 2L, 1L, 1L, 2L, 1L, 1L, 1L, 1L, 2L, 1L, 
    2L, 1L, 1L, 2L, 2L, 2L, 1L, 1L, 1L, 2L, 1L, 1L, 2L, 1L, 1L, 
    1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 1L, 1L, 2L, 1L, 2L, 
    1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
    1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L), levels = c("YES", "NO"
    ), class = "factor"), File = c("F01", "F01", "F01", "F01", 
    "F01", "F01", "F01", "F01", "F01", "F01", "F01", "F01", "F01", 
    "F01", "F01", "F01", "F01", "F01", "F01", "F01", "F01", "F01", 
    "F01", "F01", "F01", "F01", "F04", "F04", "F04", "F04", "F04", 
    "F04", "F04", "F04", "F04", "F04", "F04", "F04", "F04", "F04", 
    "F04", "F04", "F04", "F04", "F04", "F04", "F04", "F04", "F04", 
    "F04", "F04", "F04", "F04", "F04", "F04", "F04", "F04", "F04", 
    "F04", "F04", "F04", "F04", "F04", "F04", "F04", "F04", "F04", 
    "F04", "F04", "F04", "F04", "F04", "F04", "F04", "F04", "F04", 
    "F04", "F04", "F04", "F04", "F04", "F04", "F04", "F04", "F04", 
    "F04", "F04", "F04", "F04", "F04", "F07", "F07", "F07", "F07", 
    "F07", "F07", "F07", "F07", "F07", "F07", "F07", "F07", "F07", 
    "F07", "F07", "F07", "F07", "F07", "F07", "F07", "F07", "F07", 
    "F07", "F07", "F07", "F07", "F07", "F07", "F07", "F07", "F07", 
    "F07", "F07", "F07", "F07", "F07", "F07", "F07", "F07", "F07", 
    "F07", "F07", "F07", "F07", "F07", "F07", "F07", "F07", "F07", 
    "F07", "F07", "F07", "F07", "F08", "F08", "F08", "F08", "F08", 
    "F08", "F08", "F08", "F08", "F08", "F08", "F08", "F08", "F08", 
    "F08", "F08", "F08", "F08", "F08", "F08", "F08", "F08", "F08", 
    "F08", "F08", "F08", "F08", "F08", "F08", "F08", "F08", "F08", 
    "F08", "F08", "F08", "F08", "F08", "F08", "F08", "F12", "F12", 
    "F12", "F12", "F12", "F12", "F12", "F12", "F12", "F12", "F12", 
    "F12", "F12", "F12", "F12", "F12", "F12", "F12", "F12", "F12", 
    "F12", "F12", "F12", "F12", "F12", "F12", "F12", "F12", "F16", 
    "F16", "F16", "F16", "F16", "F16", "F16", "F16", "F16", "F16", 
    "F16", "F16", "F16", "F16", "F16", "F16", "F16", "F16", "F16", 
    "F18", "F18", "F18", "F18", "F18", "F18", "F18", "F18", "F18", 
    "F18", "F18", "F18", "F18", "F18", "F18", "F18", "F18", "F18", 
    "F18", "F18", "F18", "F18", "F18", "F20", "F20", "F20", "F20", 
    "F20", "F20", "F20", "F20", "F20", "F20", "F20", "F20", "F20", 
    "F20", "F20", "F20", "F20", "F20", "F20", "F22", "F22", "F22", 
    "F22", "F22", "F22", "F22", "F22", "F22", "F22", "F23", "F23", 
    "F23", "F23", "F23", "F23", "F23", "F23", "F23", "F23", "F23", 
    "F23", "F23", "F23", "F23", "F23", "F23", "F23", "F23", "F23", 
    "F23", "F23", "F23", "F23", "F23", "F23", "F23", "F23", "F23", 
    "F23", "F23", "F23", "F23", "F23", "F23", "F23", "F23", "F23", 
    "F23", "F19", "F19", "F19", "F19", "F19", "F19", "F19", "F19", 
    "F19", "F19", "F19", "F16"), Person_anon = c("GGGGGGGGGGGGGGGl", 
    "IIIIIIIIIIIIt", "IIIIIIIIIIIIt", "IIIIIIIIIIIIt", "IIIIIIIIIIIIt", 
    "IIIIIIIIIIIIt", "IIIIIIIIIIIIt", "GGGGGGGGGGGGGGGl", "IIIIIIIIIIIIt", 
    "IIIIIIIIIIIIt", "KKKKKKKKKKr", "IIIIIIIIIIIIt", "KKKKKKKKKKr", 
    "IIIIIIIIIIIIt", "IIIIIIIIIIIIt", "IIIIIIIIIIIIt", "GGGGGGGGGGGGGGGl", 
    "IIIIIIIIIIIIt", "IIIIIIIIIIIIt", "KKKKKKKKKKr", "IIIIIIIIIIIIt", 
    "IIIIIIIIIIIIt", "IIIIIIIIIIIIt", "IIIIIIIIIIIIt", "IIIIIIIIIIIIt", 
    "IIIIIIIIIIIIt", "DDDDDDDDDDDDe", "AAAAAAAAAAAAAn", "DDDDDDDDDDDDe", 
    "DDDDDDDDDDDDe", "DDDDDDDDDDDDe", "DDDDDDDDDDDDe", "DDDDDDDDDDDDe", 
    "DDDDDDDDDDDDe", "DDDDDDDDDDDDe", "CCCCCCCCCCx", "CCCCCCCCCCx", 
    "DDDDDDDDDDDDe", "DDDDDDDDDDDDe", "DDDDDDDDDDDDe", "DDDDDDDDDDDDe", 
    "DDDDDDDDDDDDe", "AAAAAAAAAAAAAn", "DDDDDDDDDDDDe", "CCCCCCCCCCx", 
    "CCCCCCCCCCx", "CCCCCCCCCCx", "AAAAAAAAAAAAAn", "AAAAAAAAAAAAAn", 
    "AAAAAAAAAAAAAn", "DDDDDDDDDDDDe", "DDDDDDDDDDDDe", "AAAAAAAAAAAAAn", 
    "DDDDDDDDDDDDe", "CCCCCCCCCCx", "DDDDDDDDDDDDe", "DDDDDDDDDDDDe", 
    "AAAAAAAAAAAAAn", "AAAAAAAAAAAAAn", "CCCCCCCCCCx", "DDDDDDDDDDDDe", 
    "AAAAAAAAAAAAAn", "CCCCCCCCCCx", "DDDDDDDDDDDDe", "DDDDDDDDDDDDe", 
    "AAAAAAAAAAAAAn", "AAAAAAAAAAAAAn", "CCCCCCCCCCx", "CCCCCCCCCCx", 
    "CCCCCCCCCCx", "DDDDDDDDDDDDe", "DDDDDDDDDDDDe", "DDDDDDDDDDDDe", 
    "DDDDDDDDDDDDe", "DDDDDDDDDDDDe", "DDDDDDDDDDDDe", "DDDDDDDDDDDDe", 
    "CCCCCCCCCCx", "DDDDDDDDDDDDe", "AAAAAAAAAAAAAn", "AAAAAAAAAAAAAn", 
    "CCCCCCCCCCx", "AAAAAAAAAAAAAn", "CCCCCCCCCCx", "DDDDDDDDDDDDe", 
    "CCCCCCCCCCx", "DDDDDDDDDDDDe", "CCCCCCCCCCx", "DDDDDDDDDDDDe", 
    "CCCCCCCCCCx", "LLLLLLLLLLLLLn", "LLLLLLLLLLLLLn", "LLLLLLLLLLLLLn", 
    "CCCCCCCCCCx", "LLLLLLLLLLLLLn", "AAAAAAAAAAAAAn", "AAAAAAAAAAAAAn", 
    "AAAAAAAAAAAAAn", "LLLLLLLLLLLLLn", "LLLLLLLLLLLLLn", "AAAAAAAAAAAAAn", 
    "AAAAAAAAAAAAAn", "CCCCCCCCCCx", "AAAAAAAAAAAAAn", "AAAAAAAAAAAAAn", 
    "AAAAAAAAAAAAAn", "CCCCCCCCCCx", "AAAAAAAAAAAAAn", "AAAAAAAAAAAAAn", 
    "AAAAAAAAAAAAAn", "CCCCCCCCCCx", "AAAAAAAAAAAAAn", "LLLLLLLLLLLLLn", 
    "LLLLLLLLLLLLLn", "AAAAAAAAAAAAAn", "AAAAAAAAAAAAAn", "LLLLLLLLLLLLLn", 
    "AAAAAAAAAAAAAn", "CCCCCCCCCCx", "CCCCCCCCCCx", "CCCCCCCCCCx", 
    "AAAAAAAAAAAAAn", "AAAAAAAAAAAAAn", "AAAAAAAAAAAAAn", "CCCCCCCCCCx", 
    "AAAAAAAAAAAAAn", "AAAAAAAAAAAAAn", "AAAAAAAAAAAAAn", "AAAAAAAAAAAAAn", 
    "CCCCCCCCCCx", "CCCCCCCCCCx", "LLLLLLLLLLLLLn", "AAAAAAAAAAAAAn", 
    "CCCCCCCCCCx", "AAAAAAAAAAAAAn", "AAAAAAAAAAAAAn", "AAAAAAAAAAAAAn", 
    "AAAAAAAAAAAAAn", "AAAAAAAAAAAAAn", "AAAAAAAAAAAAAn", "AAAAAAAAAAAAAn", 
    "AAAAAAAAAAAAAn", "LLLLLLLLLLLLLn", "LLLLLLLLLLLLLo", "LLLLLLLLLLLLLo", 
    "LLLLLLLLLLLLLn", "LLLLLLLLLLLLLn", "LLLLLLLLLLLLLo", "NNNNNNNNNNNr", 
    "NNNNNNNNNNNr", "LLLLLLLLLLLLLn", "LLLLLLLLLLLLLn", "NNNNNNNNNNNr", 
    "NNNNNNNNNNNr", "NNNNNNNNNNNr", "NNNNNNNNNNNr", "NNNNNNNNNNNr", 
    "LLLLLLLLLLLLLo", "LLLLLLLLLLLLLo", "NNNNNNNNNNNr", "LLLLLLLLLLLLLo", 
    "NNNNNNNNNNNr", "NNNNNNNNNNNr", "NNNNNNNNNNNr", "NNNNNNNNNNNr", 
    "NNNNNNNNNNNr", "LLLLLLLLLLLLLn", "LLLLLLLLLLLLLn", "NNNNNNNNNNNr", 
    "LLLLLLLLLLLLLo", "LLLLLLLLLLLLLo", "LLLLLLLLLLLLLo", "LLLLLLLLLLLLLn", 
    "LLLLLLLLLLLLLo", "NNNNNNNNNNNr", "NNNNNNNNNNNr", "LLLLLLLLLLLLLn", 
    "LLLLLLLLLLLLLn", "NNNNNNNNNNNr", "LLLLLLLLLLLLLn", "NNNNNNNNNNNr", 
    "NNNNNNNNNNNr", "LLLLLLLLLLLLLo", "LLLLLLLLLLLLLo", "CCCCCCCCCCx", 
    "LLLLLLLLLLLLLo", "DDDDDDDDDDDDe", "CCCCCCCCCCx", "LLLLLLLLLLLLLo", 
    "LLLLLLLLLLLLLo", "LLLLLLLLLLLLLo", "DDDDDDDDDDDDe", "DDDDDDDDDDDDe", 
    "DDDDDDDDDDDDe", "CCCCCCCCCCx", "DDDDDDDDDDDDe", "CCCCCCCCCCx", 
    "CCCCCCCCCCx", "CCCCCCCCCCx", "CCCCCCCCCCx", "DDDDDDDDDDDDe", 
    "CCCCCCCCCCx", "CCCCCCCCCCx", "DDDDDDDDDDDDe", "DDDDDDDDDDDDe", 
    "DDDDDDDDDDDDe", "DDDDDDDDDDDDe", "LLLLLLLLLLLLLo", "DDDDDDDDDDDDe", 
    "DDDDDDDDDDDDe", "CCCCCCCCCCx", "CCCCCCCCCCx", "AAAAAAAAAAo", 
    "AAAAAAAAAAo", "AAAAAAAAAAo", "AAAAAAAAAAo", "AAAAAAAAAAo", 
    "AAAAAAAAAAo", "CCCCCCCCCCx", "CCCCCCCCCCCCCCx", "AAAAAAAAAAo", 
    "AAAAAAAAAAo", "AAAAAAAAAAo", "AAAAAAAAAAo", "AAAAAAAAAAo", 
    "AAAAAAAAAAo", "CCCCCCCCCCCCCCx", "AAAAAAAAAAo", "CCCCCCCCCCCCCCx", 
    "SSSSSSSSSSd", "SSSSSSSSSSd", "GGGGGGGGGGGGGi", "SSSSSSSSSSd", 
    "SSSSSSSSSSd", "AAAAAAAAAAo", "SSSSSSSSSSd", "SSSSSSSSSSd", 
    "AAAAAAAAAAo", "SSSSSSSSSSd", "AAAAAAAAAAo", "SSSSSSSSSSd", 
    "SSSSSSSSSSd", "SSSSSSSSSSd", "AAAAAAAAAAo", "SSSSSSSSSSd", 
    "GGGGGGGGGGGGGi", "GGGGGGGGGGGGGi", "GGGGGGGGGGGGGi", "GGGGGGGGGGGGGi", 
    "GGGGGGGGGGGGGi", "GGGGGGGGGGGGGi", "SSSSSSSSSSd", "LLLLLLLLLLLLLLLLLLLl", 
    "LLLLLLLLLLLLLLLLLLLl", "LLLLLLLLLLLLLLLLLLLl", "LLLLLLLLLLLLLLLLLLLl", 
    "LLLLLLLLLLLLLLLLLLLl", "LLLLLLLLLLLLLLLLLLLl", "LLLLLLLLLLLLLLLLLLLl", 
    "LLLLLLLLLLLLLLLLLLLl", "LLLLLLLLLLLLLLLLLLLl", "LLLLLLLLLLLLLLLLLLLl", 
    "LLLLLLLLLLLLLLLLLLLl", "LLLLLLLLLLLLLLLLLLLl", "LLLLLLLLLLLLLLLLLLLl", 
    "LLLLLLLLLLLLLLLLLLLl", "LLLLLLLLLLLLLLLLLLLl", "LLLLLLLLLLLLLLLLLLLl", 
    "LLLLLLLLLLLLLLLLLLLl", "LLLLLLLLLLLLLLLLLLLl", "LLLLLLLLLLLLLLLLLLLl", 
    "JJJJJJJJJJJJy", "JJJJJJJJJJJJy", "JJJJJJJJJJJJy", "JJJJJJJJJJJJJJJJJJd", 
    "JJJJJJJJJJJJJJJJJJd", "JJJJJJJJJJJJJJJJJJd", "JJJJJJJJJJJJy", 
    "JJJJJJJJJJJJJJJJJJd", "JJJJJJJJJJJJy", "JJJJJJJJJJJJy", 
    "LLLLLLLLLLn", "CCCCCCCCCCCCd", "CCCCCCCCCCCCd", "CCCCCCCCCCCCd", 
    "CCCCCCCCCCCCd", "CCCCCCCCCCCCd", "OOOOOOOOOOOOm", "OOOOOOOOOOOOm", 
    "LLLLLLLLLLn", "CCCCCCCCCCCCd", "OOOOOOOOOOOOm", "CCCCCCCCCCCCd", 
    "CCCCCCCCCCCCd", "LLLLLLLLLLn", "CCCCCCCCCCCCd", "CCCCCCCCCCCCd", 
    "CCCCCCCCCCCCd", "LLLLLLLLLLn", "LLLLLLLLLLn", "LLLLLLLLLLn", 
    "OOOOOOOOOOOOm", "CCCCCCCCCCCCd", "LLLLLLLLLLn", "CCCCCCCCCCCCd", 
    "CCCCCCCCCCCCd", "OOOOOOOOOOOOm", "CCCCCCCCCCCCd", "CCCCCCCCCCCCd", 
    "CCCCCCCCCCCCd", "CCCCCCCCCCCCd", "CCCCCCCCCCCCd", "CCCCCCCCCCCCd", 
    "CCCCCCCCCCCCd", "CCCCCCCCCCCCd", "LLLLLLLLLLn", "LLLLLLLLLLn", 
    "LLLLLLLLLLn", "LLLLLLLLLLn", "OOOOOOOOOOOOm", "LLLLLLLLLLLLLLLLLLLl", 
    "LLLLLLLLLLLLLLLLLLLl", "LLLLLLLLLLLLLLLLLLLl", "LLLLLLLLLLLLLLLLLLLl", 
    "LLLLLLLLLLLLLLLLLLLl", "LLLLLLLLLLLLLLLLLLLl", "LLLLLLLLLLLLLLLLLLLl", 
    "LLLLLLLLLLLLLLLLLLLl", "LLLLLLLLLLLLLLLLLLLl", "LLLLLLLLLLLLLLLLLLLl", 
    "LLLLLLLLLLLLLLLLLLLl", "AAAAAAAAAAo")), row.names = c(NA, 
-332L), class = c("tbl_df", "tbl", "data.frame"))

Answer 1

您可以通过将您的分类结果拟合为实际的 0/1 编码虚拟效果来轻松实现此目的。这是在任何拟合例程中都会发生的情况，但后续的

predict

方法将不知道如何处理我们想要去的地方：

## This is what the design matrix looks like inside glmer()
unique(model.matrix(~N_G_altnt_Q_YN, data=df))
#>  (Intercept) N_G_altnt_Q_YN1
#>          1               0
#>          1               1

## Manually code our effect to be 0/1 (numeric)
df$X <- as.integer(df$N_G_altnt_Q_YN == 1)

## Fit this numeric effect
m_1 <- glmer(Preselected_0 ~ X + (X | File / Person_anon),
             family = "binomial", data = df)

我不会在这里展示它，但这不会改变模型的固定或随机估计中的什么，它只是从预测变量中删除了因子类型。如果你在这个因素中有两个以上的水平，你就必须创建更多的虚拟人，但同样的原则成立（同样，这就是

model.matrix

在任何模型中都会做的事情）。

最大的优势是我们现在可以

predict

0/1 以外的其他值——如果您愿意的话，甚至可以在这些范围之外，尽管这显然毫无意义。

## Helper function to back-transform log odds
expit <- function(x) 1/(1+exp(-x))

## Desired prediction range
x <- seq(0, 1, length.out = 2E2)

pred <- predict(m_1, data.frame(dummy = x), re.form = ~0, se.fit = TRUE) |>
   do.call(cbind, args = _) |>
   cbind(x) |>
   as.data.frame() |>
   dplyr::mutate(est = expit(fit),
                 ## Calculate Wald 95% confidence bounds
                 lower = expit(fit - se.fit*qnorm(.975)),
                 upper = expit(fit + se.fit*qnorm(.975)))

我们在最后一次调用中做了一些事情：询问整个预测范围内的固定效应预测平均值和标准误差，并使用这些计算 Wald 型 95% 置信区间。最后，所有内容都从对数赔率尺度反向转换为响应（概率）尺度。您需要最后执行该步骤，因为您无法直接计算响应量表中的置信区间。

现在剩下的就是制作情节了。我会坚持基本的惯例，而不是例如

ggplot2

：

plot(c(0,1), c(0,1), type="n", xlab="", ylab="", xaxt="none")
axis(1, labels=c("no", "yes"), at=0:1)
polygon(c(x, rev(x)), c(pred$upper, rev(pred$lower)), col="grey90", lty=0)
lines(pred$x, pred$est, lwd=3)

一些结束语：

• 此模型中有收敛警告。我假设这可能是虚拟数据，但这是您想要研究的内容（也许会降低随机效应的复杂性）。
• 我非常确信这样的预测是有意义的，它们将是按该比例在两种结果之间划分的总体中的平均值。显然，在个人观察层面，您只能得到两种结果之一。