|
<< Click to Display Table of Contents >> Navigation: Reference Manual > Questions > Reserve responses |
You can reserve response numbers without allocating any text or other settings to them. From the Entry details window, use the appropriate toolbar button and you will be placed in the Responses reserve.
You can choose to set "Refuse input" for any reserved responses so that these responses cannot be used.
For example, you may want response 98 to always be "Don't know". If you have a question with only 43 responses you can reserve responses 44 to 97 as "Refuse input", so that response number 98 can be used for "Don't know".
This is helpful when inputting questionnaires because the same number is always used for this response. If response lists are large in the project you may want to use 998 and in our example you would reserve 44 to 997.
An open-ended question is a special form of a multi-coded question where some or all of the responses are not known when the survey starts.
The responses are reserved in the question and as new responses are encountered they are added to the response list by allocating the new response text to one of the reserved responses.
Open-ended questions are used for lists with a limited number of potential responses, for example brands of toothpaste. In this case the main brands would be allocated to the first few responses and the rest will be reserved for those that come up during the survey.
Sometimes analysis is required from more general questions, for example "Why did you not like the product".
Each answer given can be listed but it is more useful to be able to count the number of people giving similar answers, for example "Dislike colour". Of course the actual answer could be "Horrible red", "Dirty looking", or any other text that could be interpreted as a dislike of the colour.
The name given to the grouping of answers into a limited set of responses is called "coding".
The usual way to code open-ended answers is to generate a code list (or code frame) giving all the different categories of response, allocating a response number to each.
For paper questionnaires the code list is normally generated by reading a proportion of the answer texts from the completed questionnaires and similar ones are logged using a 5 bar gate system (often grouped into positive, neutral and negative responses).
For paper projects there is usually no need to enter the full text and an open-ended question is all that is needed. After coding:
•The coded response numbers are added to the paper questionnaires
•The coded response texts are allocated to the relevant reserved responses
The data can then be entered as it would be for ordinary multi-coded questions.
TIP: we recommend you set an open-ended response allocation to a number higher than the number of responses you expect. It is easier to leave a few extra data locations when designing your questionnaire, than to increase this number once data has been input.
IMPORTANT: open-ended responses should be coded with leading zeroes as follows:
•up to 9 responses in all (width 1), enter as 1, 2, 3 and so on
•up to 99 responses in all (width 2), enter as 01, 02, 03 and so on
•up to 999 responses in all (width 3), enter as 001, 002, 003 and so on
•up to 9999 responses in all (width 4), enter as 0001, 0002, 0003, 0004 and so on
For CATI and CAPI projects the full answer text does need to be collected. This can be done with a character or verbatim question.
An open-ended question is used as well, but it is filtered so that it is not seen during the interview.
There are no paper questionnaires, so the coding is usually done in a spreadsheet and the responses are allocated to a column alongside the serial numbers. The original CSV file of verbatim questions can be used for this purpose. After coding:
•The filter is removed from the open-ended question.
•The coded response texts are allocated to the relevant reserved responses.
•The coded responses are handed-in to the data file.
If some of the responses are known before interviewing starts then an open-ended question is still used with the pre-coded responses in it and an "Other" response. The character or verbatim question is filtered so that it only appears if the "Other" response is selected.