A few months ago, Steven Broadbent kindly gave me a demonstration of Surveybe after seeing my earlier post on survey softwares for mixed method research. There were a number of features that I found quite useful. Among the highlights:
- The household roster section is designed for panel data. After a roster is created at baseline, names can still be edited and additional family members added during the follow-up. It’s also easy to enable or disable labor and education options based on household members’ ages.
- Questions can have photos displayed alongside them. My favorite part! In the sample survey, photos were used to clarify measures or definitions. For instance, the photo below is of a “medium-sized” bunch of bananas. This is ideal for agricultural surveys, where production is often measured by volume instead of weight. I imagine it would also be useful for health surveys (“what did that rash look like?”). I would have really liked this feature for the survey where I had a number of detailed conversations about which types of toilets in Kinshasa counted as “improved” – simply showing people a photo of the qualifying toilets would have been much faster.
- Support for multiple languages. Surveybe can handle non-Latin scripts, and can easily switch between instruments in different languages.
- Cool export features. Data can be exported in Stata, SPSS or CSV formats. If you’ve ever gone through multiple versions of a questionnaire trying to match variable q3_a4 to its original question, Surveybe fixes this by providing options to export question text directly to Stata. You can also set variable names within Surveybe, and they’ll remain the same even if you later change the order of the questions. (I only asked about Stata, since that’s what I use, but I assume this is also true for SPSS.)
- Data security. Exported data is automatically encrypted and can only be unencrypted by the Surveybe file manager. This sounds particularly attractive after the demise of TrueCrypt sent everyone scrambling to re-encrypt raw survey data.
Surveybe does appear to require slightly more programming competence than KoBoToolbox, particularly a minimal familiarity with SQL. Unlike most of the softwares I mentioned in my other post, it runs on Windows. There do seem to be a number of good options for Windows tablets available, and they’re comparable in price to Android tablets, which I know is a consideration for many people who end up choosing ODK-based survey softwares. While I don’t personally have plans for large-N data collection for my thesis at the moment, Surveybe will definitely be high on my list if I end up managing this type of project in the future.