R语言中执行t.test时出现‘invalid type (list) for variable’报错及数据处理问题求助

我来帮你一步步拆解和解决遇到的问题：

一、先处理Reprex里的前置错误 #

你在数据处理阶段遇到的几个错误，根源很明确：

1. object 'global' not found：你还没有创建或导入名为global的数据框，R找不到这个对象。你需要先把原始数据导入R（比如用read.csv()），或者像我下面示例里那样模拟一份测试数据。
2. no applicable method for 'tbl_vars' applied to an object of class "function"：local是R的内置函数（用来创建局部环境），不能用它作为你的数据框名称！把数据框改名成local_df这类不会冲突的名字就好。

二、核心问题：t检验的用法错误 #

你运行t.test(meansharelocal ~ meanshareglobal)时收到的invalid type (list) for variable 'meansharelocal'错误，原因是：

• meansharelocal和meanshareglobal是用summarise()得到的数据框（本质是list类型），而t.test()的公式语法（y ~ x）要求x是分组变量，y是数值向量，你不能直接把两个数据框丢进去。
• 另外，你的实验设计是两组（global和local）各有4个均值，正确的做法是把这两组均值转换成数值向量，再传入t.test()。

三、完整的修正代码示例 #

我会用模拟数据演示整个流程，你可以替换成自己的真实数据：

1. 准备数据（模拟或导入） #

library(dplyr)

# 模拟global数据框（替换成你自己的read.csv("your_global_data.csv")）
global <- tibble(
  Share_1 = sample(c("Extremely likely", "Somewhat likely", "Neither likely nor unlikely", "Somewhat unlikely", "Extremely unlikely"), 10, replace = TRUE),
  Share_2 = sample(c("Extremely likely", "Somewhat likely", "Neither likely nor unlikely", "Somewhat unlikely", "Extremely unlikely"), 10, replace = TRUE),
  Share_3 = sample(c("Extremely likely", "Somewhat likely", "Neither likely nor unlikely", "Somewhat unlikely", "Extremely unlikely"), 10, replace = TRUE),
  Share_4 = sample(c("Extremely likely", "Somewhat likely", "Neither likely nor unlikely", "Somewhat unlikely", "Extremely unlikely"), 10, replace = TRUE)
)

# 把local改成local_df，避免和内置函数冲突
local_df <- tibble(
  Share_1 = sample(c("Extremely likely", "Somewhat likely", "Neither likely nor unlikely", "Somewhat unlikely", "Extremely unlikely"), 10, replace = TRUE),
  Share_2 = sample(c("Extremely likely", "Somewhat likely", "Neither likely nor unlikely", "Somewhat unlikely", "Extremely unlikely"), 10, replace = TRUE),
  Share_3 = sample(c("Extremely likely", "Somewhat likely", "Neither likely nor unlikely", "Somewhat unlikely", "Extremely unlikely"), 10, replace = TRUE),
  Share_4 = sample(c("Extremely likely", "Somewhat likely", "Neither likely nor unlikely", "Somewhat unlikely", "Extremely unlikely"), 10, replace = TRUE)
)

2. 重新编码分类变量为数值 #

用across()替代旧的mutate_at（dplyr 1.0.0版本后推荐用法）：

globalshare <- global %>%
  mutate(across(c(Share_1, Share_2, Share_3, Share_4), 
                ~recode(., 
                        "Extremely likely" = 4, 
                        "Somewhat likely" = 3, 
                        "Neither likely nor unlikely" = 2, 
                        "Somewhat unlikely" = 1, 
                        "Extremely unlikely" = 0)))

localshare <- local_df %>%
  mutate(across(c(Share_1, Share_2, Share_3, Share_4), 
                ~recode(., 
                        "Extremely likely" = 4, 
                        "Somewhat likely" = 3, 
                        "Neither likely nor unlikely" = 2, 
                        "Somewhat unlikely" = 1, 
                        "Extremely unlikely" = 0)))

3. 计算均值并转换为数值向量 #

# 提取均值并转成数值向量，而不是保留数据框格式
meanshareglobal <- as.numeric(globalshare %>% summarise(mean(Share_1), mean(Share_2), mean(Share_3), mean(Share_4)))
meansharelocal <- as.numeric(localshare %>% summarise(mean(Share_1), mean(Share_2), mean(Share_3), mean(Share_4)))

4. 执行t检验 #

根据你的实验设计选择合适的检验方式：

• 独立样本t检验（假设4个均值是独立观测）：

  t.test(meanshareglobal, meansharelocal)
  

• 配对t检验（如果Share_1到Share_4是对应的类别，比如四个相关问题，适合配对比较）：

  t.test(meanshareglobal, meansharelocal, paired = TRUE)
  

四、额外提示 #

如果你想直接用原始的Share变量做检验（而不是先算均值），可以把数据转成长格式后再用公式语法，比如：

# 把global和local数据合并并标记分组
combined_data <- bind_rows(
  globalshare %>% pivot_longer(cols = everything(), names_to = "Share", values_to = "Score") %>% mutate(Group = "Global"),
  localshare %>% pivot_longer(cols = everything(), names_to = "Share", values_to = "Score") %>% mutate(Group = "Local")
)

# 用公式语法做t检验
t.test(Score ~ Group, data = combined_data)

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