There are a variety of ways of exporting data from Alida's Sparq platform. It is believed that the best method is as follows, in Sparq:
- Click on Dynamic Reporting > Quick Reports (top of the screen).
- Press the Export button.
- Export > New Export and select SPSS (Data Export).
- Set Select Data to Live Data and Complete Only.
- Set Select questions to All question types.
- Select all of the questions that you desire to export (typically this will be all of them).
- Press Export.
- Press Download and then Save.
- Move the file to somewhere so that you can open it in Displayr.
- Import the data into Dispayr and fix up the problems (see below), following the steps in Setting Up Files With No Metadata.
Users of Sparq have reported the following issues. In general, these need to be manually fixed once the data has been imported into Displayr.
Variable Labels for grids and loops
SPSS data files decompose each question in a survey into one or more variables (the data files do not really have a concept of a question). Each variable has metadata associated with it. One piece of metadata is called the Variable Label. Displayr uses the contents of the Variable Label to deduce the existence of multiple response questions, grid questions, looped questions, and question-wording, as none of this information is explicitly contained within an SPSS data file. For example, in a good data file the labels of variables for a grid question would be as follows:
Q11. Please rate your satisfaction with: Online Banking – Bank of America Q11. Please rate your satisfaction with: Online Banking – CitiBank Q11. Please rate your satisfaction with: Mobile Banking – Bank of America Q11. Please rate your satisfaction with: Mobile Banking – CitiBank
Such labels are informative, as from them it can be determined that:
- A question was asked with a name of Q11. which asked Please rate your satisfaction with
- The question was a type of grid, in which people were permitted to choose multiple options in rows and columns (i.e., in Displayr jargon, a Binary - Grid).
- The names of the rows or columns were: Online Banking, Mobile Banking and Bank of America, Bank of America
When such data is outputted from Sparq, the contents of the labels can be as terse as:
Q11a Q11b Q11c Q11d
Thus, there is no way to deduce from the data file what questions were asked, and it is necessary to go into Sparq to figure it out and then enter the correct information into Displayr.
Sparq files typically do not contain:
- A time and date stamp for interview commencement and completion. This is used both for quality control purposes and also to permit time series analyses.
- A unique identifier variable for the interview (i.e., essentially a unique key), so that when data is cleaned it is possible to repeat this cleaning in revised data files.
Non-stacking of waves of trackers
When a member of the panel completes a questionnaire on multiple occasions, the additional data is stored as additional variables. This means that it has to be stacked (see Stacking Data Files).
Incorrect Measure Type
Another type of metadata in SPSS files is the Measure, which is often set in Sparq files at Nominal when it should be set as Scale (which causes Displayr to treat data as categorical that is in fact numerical).