Open-Ended Responses: Tell Me What You Think
We had a survey looking at brand loyalty, but were unclear on what the fundamental reason for brand loyalty was. As such, we simply left it as an open ended option and let people tell us.
If you believe the responses are going to be widely varied and you don’t want to bias people toward a particular set, leave the responses open-ended.
If you want to refine a survey over time, you can include an open-ended response as an option along with a predefined list. Eventually, you might want to remove options that have a low response rate and replace them with those that have higher rates among your open ended responses.
Re-coding open-ended responses
Managing open-ended re-codes at the end of a project is an interesting process on its own.
Re-coding is tedious, but important, and requires the full attention of whoever is working on it.
Because the process is subjective, confirming accuracy can be difficult. I would recommend, if you have the time and capability, having two different individuals re-code. So long as your results are not significantly different you can have a higher level of confidence that the re-coding is accurate.
Spot checking the data can also help to reassure your confidence.
Tracing open-ends back to their respective questions
It is also highly important that the data is able to be quickly identified.
I recommend making each question its own page, with a respondent ID tagging each response on each page.
This, combined with Excel’s VLOOKUP function, gives you an easy way to add your results back into the survey, should you wish to analyze them in that way.
It is often appropriate to treat your questions as multi-response, as people have a tendency to supply a bevy of responses in open ended boxes. This also prevents you from having to condense useful results into vague categories, or missing out on including some responses.
Photo Source: https://www.flickr.com/photos/gbsk/