Quantifying truncation errors in effective field theory