CODEWITHSUNDEEP
rsurvey
filups21
filups21

Reputation: 1917

R survey svymean returns 0 with no data

Why does the survey package in R return a mean value of 0 when calculating the mean of no data? Shouldn't the result be NA?

Example:

survey::svymean(
   ~x, 
   design = survey::svydesign(
      ids = ~1, 
      weights = ~w, 
      data = data.frame(
         x = rep(NA_real_, 100),
         w = rep(1,100)
      )
   ), 
   na.rm = TRUE
)

returns:
mean SE
x 0 0


Without the na.rm=TRUE, the function returns NA:

survey::svymean(
   ~x, 
   design = survey::svydesign(
      ids = ~1, 
      weights = ~w, 
      data = data.frame(
         x = rep(NA_real_, 100),
         w = rep(1,100)
      )
   )
)

returns:
mean SE
x NA NaN

Upvotes: 0

Views: 194

Answers (1)

Thomas Lumley
Thomas Lumley

Reputation: 2765

Because that's what the code happens to return in this case and it wasn't thought to be worth adding a special case for the combination of no non-missing data and na.rm=TRUE

Upvotes: 1

Related Questions