如何调整tab_model输出的MLM系数顺序并指定显示项？

核心问题原因 #

你设置的auto.label = TRUE会覆盖自定义的系数顺序和标签，这个参数会自动从模型中提取变量标签和默认顺序，导致你指定的terms和pred.labels失效。

具体解决步骤 #

• 关闭自动标签功能：把auto.label = TRUE改成auto.label = FALSE，这样tab_model才会遵循你指定的顺序和自定义标签。
• 对齐pred.labels和terms的内容：你的coefs里包含了mother_dummy_12011，但terms里没列这个变量，建议把pred.labels调整为只包含terms里的变量，且顺序和terms完全一致，避免标签错乱。
• （可选）提前设置因子顺序：如果mother_l1_recode是因子变量，在拟合模型前就用factor()函数指定因子水平顺序，比如：

  df$mother_l1_recode <- factor(df$mother_l1_recode, 
                                levels = c("mother_l1_recode1", "mother_l1_recode2", ..., "mother_l1_recodenot_provided"))
  

  这样模型拟合后，默认输出顺序就是你想要的，后续用tab_model时更省心。

修改后的代码示例 #

# 调整pred.labels，只保留terms里的变量，顺序和terms一致
coefs <- c(
  "mother_l1_recode1" = "Northern European",
  "mother_l1_recode2" = "Southern European",
  "mother_l1_recode3" = "Eastern European",
  "mother_l1_recode4" = "Southwest and Central Asian",
  "mother_l1_recode5" = "Southern Asian",
  "mother_l1_recode6" = "Southeast Asian",
  "mother_l1_recode7" = "Eastern Asian",
  "mother_l1_recode9" = "Other",
  "mother_l1_recodenot_provided" = "Not provided"
)

tab_model(mlm1, mlm1_non1201,
          title = " ",
          pred.labels = coefs,
          terms = c(
            "mother_l1_recode1", 
            "mother_l1_recode2",
            "mother_l1_recode3", 
            "mother_l1_recode4",
            "mother_l1_recode5", 
            "mother_l1_recode6",
            "mother_l1_recode7", 
            "mother_l1_recode9",
            "mother_l1_recodenot_provided"
          ),
          show.intercept = FALSE,
          show.ci = FALSE,
          show.se = TRUE,
          string.se = "S.E.",
          dv.labels = c(
            paste("Year", parse_number(Year_Level_2_use), sep = " "),
            paste("Year", parse_number(Year_Level_2_use)+4, sep = " ")
          ),
          string.pred = as.character(year),
          string.est = "Coefficient",
          p.style = "stars", 
          show.icc = FALSE,
          show.re.var = FALSE,
          show.obs = FALSE,
          show.r2 = FALSE,
          show.ngroups = FALSE,
          auto.label = FALSE  # 关键修改：关闭自动标签
)

内容的提问来源于stack exchange，提问作者m45ha