dplyr mutate() or transmute() column gets nested in data frame

[bayesian]

I am trying to calculate a survey scale score which involves subtracting from certain items. However, I am not sure how to properly add my end result back to my data frame as a proper column. Checking str(df) I realise the computed column is nested.

# for reprex 
df <- structure(list(column0 = c(10L, 10L, 10L), 
                     column1 = c(4L, 3L, 4L), 
                     column2 = c(2L, 3L, 4L), 
                     column3 = c(1L, 2L, 1L),
                     blah = c(1L, 2L, 1L)), 
                class = "data.frame", 
                row.names = c(NA, -3L))
df

  column0 column1 column2 column3 blah
1      10       4       2       1    1
2      10       3       3       2    2
3      10       4       4       1    1

# columns to subtract from
subtract.cols <- c("column1", "column2", "column3")

# sum all columns that contain "column" in the name, except subtract.cols
# store the result as column5
df$column5 <- df %>% select(contains("column")) %>% 
               mutate(hahaha = rowSums(.) - rowSums(.[subtract.cols]) ) %>% 
               select("hahaha")

Calling column5 gives me the column named hahaha

head(df %>% select("column5"))

  hahaha
1     10
2     10
3     10

Apparently because hahaha is nested within a the df column5

str(df)

'data.frame':	3 obs. of  6 variables:
 $ column0: int  10 10 10
 $ column1: int  4 3 4
 $ column2: int  2 3 4
 $ column3: int  1 2 1
 $ blah   : int  1 2 1
 $ column5:'data.frame':	3 obs. of  1 variable:
  ..$ hahaha: num  10 10 10

How can I get the computed results in a normal column like the rest of my data?

[nirgrahamuk]

@bayesian

df$column5 <- df %>% select(contains("column")) %>% 
               mutate(hahaha = rowSums(.) - rowSums(.[subtract.cols]) ) %>% 
               select("hahaha")

Change the last select to be pull