The non-sampling errors, such as due to measuring and recording observations, inaccuracy or incompleteness of information, location of units, non-response or incomplete response, training of investigators, interpretation of questions, bias of the investigators, etc., are of a more serious nature in a complete census.
In a sample survey, these errors can be effectively controlled and minimized by (i) employing highly qualified, skilled and trained personnel, (ii) imparting adequate training to the investigations for conducting the enquiry, (iii) better supervision, (iv) using more sophisticated equipment and statistical techniques for the processing and analysis of the relatively limited data.