The pre-codes that are used on the questionnaire determine what data are collected. If the pre-codes have insufficient accuracy or are incomplete, then data will be lost that may be important to answering the objectives. In many instances the responses will be obvious – yes-no, male-female – but in others care must be taken to ensure that they are:
- mutually exclusive;
- as exhaustive as possible;
- as precise as necessary;
- meaningful.
Unless they are mutually exclusive, it will be possible to code the same response against more than one response code. This is confusing for the interviewer (or respondent with a self-completion questionnaire) and makes the output ambiguous and impossible to interpret (see Figure 7.1).
The pre-codes need to be as exhaustive as they can in order to minimize the number of ‘other answers’ written in. If there are a lot of ‘other answers’ written in, the question would better have been recorded as an open-ended one.
1. Recording values
When recording answers that are values, the level of detail needs to be as precise as is necessary to meet the research objectives without demanding more detail than respondents can accurately give. Sometimes it is possible to record precise values (eg the number of times the respondent has visited a pub or bar in the past week), but frequently we do not want to record that level of detail, and nor can respondents be expected to provide it. Then the answers will be recorded in value bands.
In Figure 7.2, the questionnaire writer has determined that bands of £200 are sufficiently accurate to meet the demands of the study. Bands of £50 would have given the researcher greater accuracy in calculating the average cost of a holiday and in making comparisons between sub-groups, but might have been difficult for respondents to recall accurately. This could have led to an increase in the proportion of ‘Don’t know’ responses. In this example the response categories are exclusive. If someone had paid exactly £400, it is clear where the answer should be coded.
The pre-code response categories must also be meaningful to both respondent and researcher if the first is to be able to answer and the second to interpret. Precise wording is important in achieving clarity. Words such as ‘often’, ‘frequently’ and ‘occasionally’ are best avoided, as their interpretation varies between situations and between people.
2. Constructing ranges
Wherever possible, values should be recorded as absolute numbers (but beware of duplication, as illustrated in Figure 7.3). However, if values are to be recorded in ranges, the ranges should usually be constructed such that the most popular values occur in the middle of the ranges. For example, if the question is ‘How much did you pay for the paperback novel that you are currently reading?’, we know that most answers, if accurately given, will be £x.99. However, it would not be unusual to see the following ranges given for this question:
This can cause loss of accuracy. A book costing £6.99 will be reported by some respondents as costing that amount precisely. Other respondents will round it up to £7, and the response will be recorded in the category above the one it should be in. Other respondents may say ‘about £7’, leaving the interviewer unsure as to where it should be coded. As importantly, in the analysis of these data we may want to produce an average price paid. Having collected the data in these ranges, we would normally allocate the value of the mid-point of each range to calculate the average. However, if nearly all of the actual values are at the top end of each range, the calculated average price paid will be around 50p below what it should be.
Source: Brace Ian (2018), Questionnaire Design: How to Plan, Structure and Write Survey Material for Effective Market Research, Kogan Page; 4th edition.
20 Aug 2021
20 Aug 2021
20 Aug 2021
20 Aug 2021
20 Aug 2021
20 Aug 2021