Editing eliminates errors in MR Data

Editing is the step whereby researchers eliminate errors or points of confusion in the raw data. Coding is the step whereby researchers assign respondents answers to pertinent response categories thereby expediting the tabulation. These steps are frequently accomplished by the same individuals and often in the same operation.

There are three important reasons for studying these topics:

1. The effectiveness of the entire analysts function may be hampered because of poor editing and coding.
2. Editing helps researchers evaluate the filed force, the effectiveness of the questionnaire, and the survey operations in general. This can provide information about how the research procedure can be improved in the future.
3. Since editing and coding costs may exceed 25 percent of the total survey costs, researchers will want to make sure that both of these steps are properly carried out.

Planning the editing and coding: The persons who are to do the editing and coding work should be familiar with the field interviewing procedures. If at all possible, they should be exposed to the interviewer training program and participate in the interviewer control work.

The design of the questionnaire should anticipate the editing and coding work, since the physical arrangement of the data collection form must allow space for editing and coding. If the data are to be computer tabulated, codes are assigned to the alternative response categories whenever possible and included on the questionnaire itself. This is referred to as pre-coding. In the yogurt example presented the fourth question inquired about the household’s annual income. The researchers used six responses categories for this question and pre-coded them (1) through (6) as follows:

—–Less than $15,000 (1) ——- $35,000 or more (4)
—–$15,000 – 24,999 (2) ——- Don’t know (5)
—– $25,500 – 34,999 (3) ——- Refuse /no answer

Editing Procedures: These should be in writing and should explain in detail how the answers to each question are to be handled. These instructions are prepared only after a sizable number of questionnaires have been studied and considerable though has given to what tabulations will be required. Examples of how to edit the raw data should be included.

Example: One survey asked respondents whether they bought a majority of their groceries at a chain store or at an independent store. Later in the questionnaire, they were asked the name of the grocery store they patronized “more than any other”. A number of respondents who had answered earlier that they bought primarily from a chain store gave the name of an independent store. Editors were instructed to change the “chain store” answers to independent store answers whenever they encountered such questionnaires.

The instruction or handling difficult questions (e.g. open questions) should be discussed in detail and illustrated with examples taken from the particular study. Following this, the editors “practice” on a sample of a questionnaires and their work is reviewed critically the results of the practice work indicate whether it is necessary to revise the editing and coding instruction anticipate all the problems that will arise. Therefore, the survey leader must maintain constant contact with the work in progress.

All editing is done using a writing instrument with a color other than the one used by the interviewers. Unless this is done there is no way to distinguish between “original” and “edited” data. Editors must not destroy original data by erasure. Rather, where necessary original entries are deleted by drawing a light line through them.

Defecting Incorrect Answers: It is sometimes possible to detect incorrect answers when to two or more questions are inconsistent he handling of such answers depends upon the nature of the inconsistency.

Example: A survey was conducted to find out what percent of those families owning a microwave oven had purchased it for less than list price. After finding out whether a microwave oven was in the home, the interviewer asked the brand and model. The answers were verified by observation. A later question asking the price at which the appliances had been purchased resulted in a number of answers that indicated a price substantially higher than the known tail list price. Such answers were deleted by the editors, for the answer was obviously incorrect and there was no way the correct answer could be “edited in”.