Question

我已使用summary()包中的mediation函数，并希望将结果转换为data.frame（因此我可以在将其保存为CSV之前进行更多操作文件）。

这是summary(mediation)输出的示例。

我在tidy包和普通broom函数中尝试了data.frame()函数但没有成功。

Answer 1

更新（忘记下面的丑陋的事情）

extract_mediation_summary <- function (x) { 

  clp <- 100 * x$conf.level
  isLinear.y <- ((class(x$model.y)[1] %in% c("lm", "rq")) || 
                   (inherits(x$model.y, "glm") && x$model.y$family$family == 
                      "gaussian" && x$model.y$family$link == "identity") || 
                   (inherits(x$model.y, "survreg") && x$model.y$dist == 
                      "gaussian"))

  printone <- !x$INT && isLinear.y

  if (printone) {

    smat <- c(x$d1, x$d1.ci, x$d1.p)
    smat <- rbind(smat, c(x$z0, x$z0.ci, x$z0.p))
    smat <- rbind(smat, c(x$tau.coef, x$tau.ci, x$tau.p))
    smat <- rbind(smat, c(x$n0, x$n0.ci, x$n0.p))

    rownames(smat) <- c("ACME", "ADE", "Total Effect", "Prop. Mediated")

  } else {
    smat <- c(x$d0, x$d0.ci, x$d0.p)
    smat <- rbind(smat, c(x$d1, x$d1.ci, x$d1.p))
    smat <- rbind(smat, c(x$z0, x$z0.ci, x$z0.p))
    smat <- rbind(smat, c(x$z1, x$z1.ci, x$z1.p))
    smat <- rbind(smat, c(x$tau.coef, x$tau.ci, x$tau.p))
    smat <- rbind(smat, c(x$n0, x$n0.ci, x$n0.p))
    smat <- rbind(smat, c(x$n1, x$n1.ci, x$n1.p))
    smat <- rbind(smat, c(x$d.avg, x$d.avg.ci, x$d.avg.p))
    smat <- rbind(smat, c(x$z.avg, x$z.avg.ci, x$z.avg.p))
    smat <- rbind(smat, c(x$n.avg, x$n.avg.ci, x$n.avg.p))

    rownames(smat) <- c("ACME (control)", "ACME (treated)", 
                        "ADE (control)", "ADE (treated)", "Total Effect", 
                        "Prop. Mediated (control)", "Prop. Mediated (treated)", 
                        "ACME (average)", "ADE (average)", "Prop. Mediated (average)")

  }

  colnames(smat) <- c("Estimate", paste(clp, "% CI Lower", sep = ""), 
                      paste(clp, "% CI Upper", sep = ""), "p-value")
  smat

}

并且，在行动中：

extract_mediation_summary(summary(med.out))
##                             Estimate 95% CI Lower 95% CI Upper p-value
## ACME (control)           0.083271207   0.03938198    0.1334100    0.00
## ACME (treated)           0.083228279   0.03685293    0.1328788    0.00
## ADE (control)            0.009957454  -0.10992815    0.1705339    0.90
## ADE (treated)            0.009914526  -0.12182752    0.1781260    0.90
## Total Effect             0.093185734  -0.03049831    0.2458085    0.16
## Prop. Mediated (control) 0.807209060  -7.10847324    8.8136165    0.16
## Prop. Mediated (treated) 0.820068951  -6.45663184    8.2852297    0.16
## ACME (average)           0.083249743   0.03910592    0.1322914    0.00
## ADE (average)            0.009935990  -0.11587783    0.1743300    0.90
## Prop. Mediated (average) 0.813639005  -6.78255254    8.5494231    0.16

丑陋，但确实有效：

library(mediation)
library(sandwich)

data("framing")
med.fit <- lm(emo ~ treat + age + educ + gender + income, data = framing)
out.fit <- glm(cong_mesg ~ emo + treat + age + educ + gender + income,
               data = framing, family = binomial("probit"))


med.out <- mediate(med.fit, out.fit, treat = "treat", mediator = "emo",
                   robustSE = TRUE, sims = 100)

# extract summary

library(purrr)

summary(med.out) %>%
  capture.output() %>%
  discard(`==`, "") -> lines

lines[which(grepl("^ ", lines)):(which(grepl("^Sample", lines))-1)] %>%
  sub("^       ", "med.out", .) %>%
  gsub(" 95%", "_95%", .) %>%
  gsub("CI ", "ci_", .) %>%
  sub(" \\(", "_(", .) %>%
  sub("p-", "p_", .) %>%
  sub(" ", "_", ., fixed=TRUE)  %>%
  textConnection() %>%
  read.table(header=TRUE) %>%
  setNames(sub("_$", "", colnames(.))) %>%
  dplyr::mutate(med.out=sub("\\.|_$", "", med.out),
                med.out=gsub("_", " ", med.out))
##                    med.out Estimate_95. ci_Lower_95. ci_Upper p_value
## 1           ACME (control)       0.0873       0.0382   0.1593    0.00
## 2           ACME (treated)       0.0876       0.0434   0.1586    0.00
## 3            ADE (control)       0.0115      -0.0869   0.1169    0.86
## 4            ADE (treated)       0.0117      -0.0986   0.1238    0.86
## 5             Total Effect       0.0991      -0.0118   0.2160    0.10
## 6  Prop Mediated (control)       0.7862      -4.2247   6.1448    0.10
## 7  Prop Mediated (treated)       0.8044      -3.7360   5.6117    0.10
## 8           ACME (average)       0.0875       0.0409   0.1556    0.00
## 9            ADE (average)       0.0116      -0.0932   0.1204    0.86
## 10 Prop Mediated (average)       0.7953      -3.9803   5.8783    0.10

如何将R中介摘要转换为data.frame？

1 个答案: