如何使用tbl_svysummary确保舍入均匀

Question

我正在尝试使用

survey

对象生成描述性统计表

gtsummary

包。我想根据受访者是否投票给特朗普来纳入描述性统计数据的差异。我不知道如何控制

add_differences()

产生的位数，以便它与表的其余部分匹配。

这是一些示例代码：

# Load necessary packages
library(survey)
library(dplyr)
library(gtsummary)

# Create a sample data frame with a weight variable
set.seed(123)
# Create a sample data frame
set.seed(123)
df <- data.frame(
  age_rs = runif(100, 18, 80),  # rescaled age variable
  income_by_20_rs = runif(100, 0, 1),  # rescaled income variable
  physical_health_poor = runif(100, 0, 1),  # continuous physical health variable
  all_religion1 = runif(100, 0, 1),  # continuous religious status variable
  racial_consciousness = runif(100, 0, 1),  # rescaled racial consciousness
  number_discrim_areas_rs = runif(100, 0, 5),  # number of discrimination areas (rescaled)
  vote_trump = factor(sample(c(0, 1), 100, replace = TRUE), labels = c("No", "Yes")),
  weights = runif(100, 0.5, 2)  # weight variable ranging between 0.5 and 2
)

# Create survey design object with weights
df_design <- svydesign(ids = ~1, data = df, weights = ~weights)

# Make table using gtsummary 
tbl1 <- tbl_svysummary(
  data = df_design, 
  by = "vote_trump", 
  statistic = list(all_continuous() ~ "{mean} ({sd})"), 
  include = c(age_rs, income_by_20_rs, physical_health_poor, all_religion1),
  label = list(age_rs = "Age", 
               income_by_20_rs = "Income", 
               physical_health_poor = "Poor Physical Health", 
               all_religion1 = "Religious Devotion"),
  digits = list(all_continuous() ~ c(3,3)),
  missing = "no",
  type = list(c(age_rs, income_by_20_rs, physical_health_poor, all_religion1) ~ "continuous") 
  ) %>% 
  add_difference() 

tbl1

不幸的是，我不知道如何舍入差值和 p 值列，以便它们与描述性统计数据匹配。我也想去掉 CI 栏。

下面是表格的图片：

Answer 1

您可以使用

add_difference(estimate_fun)

和

add_difference(pvalue_fun)

参数来更改对估计值和 p 值进行舍入/格式化的函数。最后，您可以使用

modify_column_hide()

隐藏打印的列。

library(gtsummary)

# Create a sample data frame with a weight variable
set.seed(123)
df <- data.frame(
  age_rs = runif(100, 18, 80),  # rescaled age variable
  income_by_20_rs = runif(100, 0, 1),  # rescaled income variable
  physical_health_poor = runif(100, 0, 1),  # continuous physical health variable
  all_religion1 = runif(100, 0, 1),  # continuous religious status variable
  racial_consciousness = runif(100, 0, 1),  # rescaled racial consciousness
  number_discrim_areas_rs = runif(100, 0, 5),  # number of discrimination areas (rescaled)
  vote_trump = factor(sample(c(0, 1), 100, replace = TRUE), labels = c("No", "Yes")),
  weights = runif(100, 0.5, 2)  # weight variable ranging between 0.5 and 2
)

# Create survey design object with weights
df_design <- survey::svydesign(ids = ~1, data = df, weights = ~weights)

# Make table using gtsummary 
tbl1 <- tbl_svysummary(
  data = df_design, 
  by = "vote_trump", 
  statistic = list(all_continuous() ~ "{mean} ({sd})"), 
  include = c(age_rs, income_by_20_rs, physical_health_poor, all_religion1),
  digits = list(all_continuous() ~ c(3,3)),
  missing = "no",
) |> 
  add_difference(
    # round estimates to 3 decimal places
    estimate_fun = everything() ~ label_style_number(digits = 3),
    # round pvalues to 3 decimal places
    pvalue_fun = label_style_pvalue(digits = 3)
  ) |> 
  # hide the CI: you can run `show_header_names(tbl)` to see the underlying column names
  modify_column_hide(columns = c(conf.low, conf.high))

as_kable(tbl1) # convert to kable to display on stackoverflow

特点	否 N = 56	是 N = 67	区别	p 值
年龄_rs	48.274 (17.960)	50.115 (17.450)	1.842	0.625
收入_by_20_rs	0.563 (0.299)	0.481 (0.242)	-0.082	0.171
身体健康状况不佳	0.437 (0.276)	0.519 (0.302)	0.082	0.179
所有_宗教1	0.449 (0.279)	0.536 (0.297)	0.088	0.151

^{创建于 2024 年 12 月 30 日，使用 reprex v2.1.1}