Coding is the procedure of classifying the answers to a question into meaningful categories The symbols used to indicate these categories are called codes. Coding is necessary to carry out the subsequent operations of tabulating and analyzing data. If coding is not done, it will not be possible to reduce a large number of heterogeneous responses into meaningful categories with the result that the analysis of data would be weak and ineffective, and without proper focus.
Coding involves two steps. The first step is to specify the different categories or classes into which the responses are to be classified. The second step is to allocate individual answers to different categories.
Code construction, as pointed out by Sidel, is something of an art and the final categories reflect the tastes and interests of the individual researcher. All the same, it is desirable to follow some guidelines to ensure the utility and rationality of the code. One of the most important points in this respect is that the categories must be all inclusive and mutually exclusive. The ‘all-inclusive’ aspect can be taken care of by adding one or more such categories as ‘other’, ‘no information’ and ‘none’. The other aspect is that categories must be ‘mutually exclusive’, i.e., they must not be overlapping and ambiguous. It should be possible to classify each response in one and only one category. However this requirement is often violated when more than one dimension is embodied in a single case. To give an example, a persons may, by occupation, be an industrial worker as well as unemployed. Here, two concepts or dimensions have been used. The first is the occupational category and the second is the current employment status. In such a case, there is apprehension that different categories or classes will not be mutually exclusive. It would, therefore, be advisable to use two category-sets, one for the occupations and the other for the current employment status.
There is no definite rule for the number of categories or classes that can be used. This will depend on the research problem as also the extent of analysis the researcher proposes to carry out. In large surveys, where mostly structured questionnaires are used, the response categories are pre-determined and are contained in the questionnaires themselves. The categories are in the form of multiple-choice answers to the question. For example, the respondent may be asked: To which age group do you belong? 15-30 years, 30-45 years, 45-60 +years. Here, four distinct categories are indicated and the respondent is supposed to indicate by checking the category in which his age falls. It is obvious that in such a case the respondent himself chooses the category which is applicable to him.
The problem of coding is not so simple, especially in respect of an open ended question. The response to such a question is in a descriptive form, in the words of the respondent himself. For example, the respondent may be asked- What is your opinion regarding the prohibition policy of the government? The respondent may give a lengthy answer indicating what he feels about this policy. In case of such responses, coding needs extra care in framing the possible categories in which various responses can be classified. Sometimes the interviewer himself.
| Questions | Answers | Codes |
| How often these days do you go to Cinema? | More than once a week Once a week Once a fortnight Once a month Three or four times a year | 1 2 3 4 5 6 |
| Less often | Never | 7 |
| Which type of wrist watch | Hand wound | 1 |
| Do you own? | Automatic | 2 |
| Which of the following Battery operated equipment Battery operated equipment | Electronic Torch Transistor Others (specify) | 3 1 2 3 |
Decides the category in which a particular response to an open-ended question is to be included. He may first take down the entire response and then decide the category in which it should be included.
At times the questionnaires are wholly or partially pre-coded. These questionnaires contain a numeric code for each of the response categories. Sometimes data are transferred from the questionnaire to a coding sheet and then punched into cards.
A practice which is frequently followed is to edit and code the date simultaneously. These two operations are regarded as one operation which is looked after by one person. As has been rightly pointed out, although this may perhaps be the quickest and most efficient method, it may lead to the neglect of editing as the editor who is expected to code becomes just a coder. In view of this, it may be advisable to get these jobs done by two persons. However, in such a case, coding by itself tends to be monotonous and bring but this perhaps cannot be helped.
