I am keeping raw survey data structured so that compass does not automatically close loops. This was excellent advice I received from Jim Coke, to prevent the data and errors from being distorted.
When the line survey becomes a cartographic survey we will worry about the error correction. By using this approach its very easy to visualize the degree of error. Compass gives error reports on closed loops, but its definitely better to see the data exactly as you surveyed it. I then measure closure errors with the measure tool.
The question here is what degree of error do surveyors here consider acceptable before feeling the need to resurvey? All input very much appreciated.