Moving from interviews to assignments
Conceptually, version 5.21 introduces an important change in how interview assignments are distributed.
Previously, survey managers had a binary choice for distributing assignments: sample or census mode. Sample mode offered complete control over survey targets, but imposed the cost of specifying all identifying information for each survey target (e.g., name, address, etc.). Census mode offered complete flexibility for identifying survey targets, but provided no control over the identifying information of potential survey targets (e.g., region where households have been selected).
Now, with assignments, survey managers have a continuum of choices: sample mode, census mode, and everything in between. To be clear, users can still distribute interviews just as before. But, importantly, there are also new choices that are more closely aligned with common, but previously unmet, user needs. To see this, consider a few use cases. When faced with HQ-based selection of primary sampling units (PSUs) but field-based sampling of households, survey managers can now pre-populate identifying details of PSUs (e.g., region, district, village name) while requiring interviewers to provide identifying details of households selected for interview (e.g., household head's name, etc.). When confronted with tracking members of a panel study, survey managers can now dynamically determined the number of copies of interviews associated with an assignment, so that any split-offs from an original household can be quickly captured with the benefit of any preloaded information.
Practically, this conceptual change in version 5.21 is relatively minor for all users.
For Headquarters, there is no longer any need to choose the mode of distribution (i.e., select sample or census mode). When assignments are created, the mode of distribution is determined by the quantity of interviews associated with an assignment (i.e., 1 to replicate sample mode, unlimited to replicate census mode, and for new modes some number in between).
For Supervisor, survey management now happens in two spaces. When making or revising assignments, supervisors work from the Assignments screen. When reviewing completed interviews, supervisors work from the Interviews screen. In other words, there is a space for each entity. Assignments in Assignments; Interviews in Interviews.
For Interviewer, there is a new tab on the dashboard: Create New. From this tab, interviewers can create interviews from their assignments. For example, if an assignment consists of three interviews, the interviewer can create up to three interviews. Once an interview is created, it moves from one status tab to another as before (e.g., Started, Completed, etc.). Meanwhile, the list of all assignments will remain in the Create New tab.
For a compact but comprehensive summary of what's new and how legacy questionnaires are handled, read here.
The Interviewer application has seen three improvements aimed at increasing usability.
- Interviewer app updates are incremental. Previously, updating the Interviewer application required downloading the full installation file with each new release. Now, updating the application downloads a smaller file, making it easier and less costly for interviewers to keep their tablets up to date in the field. Upgrading from the previous version to the current version might require downloading just few hundred kilobytes.
- Dashboard cards are expandable. Previously, assignment cards displayed only the first three identifying questions. Now, cards can be expanded to see all identifying questions. To expand a card, simply tap. To collapse, tap the card again.
- Dashboard interface uses swipe navigation. Previously, interviewers needed to tap on the status tab to move between lists of interviews (e.g., Started, Completed). Now, they can swipe right or left.
The Designer application has also seen several improvements to usability:
- Login with email address. Previously, users needed to enter their login in order to access Designer or recover their password. Now, they may also use the email address used to sign up for their Designer account, making login and password recovery easier.
- Save partially complete questions and validations. Previously, questionnaire designers needed to complete the question text field before saving a question, and provide an error message to save a validation. Now, designers may leave both blank. Questions without question text will appear as compilation errors. Validations without error messages will simply not show an error message for invalid answers. These changes makes creating rough drafts easier.
- Simplifies data types. Previously, Survey Solutions several C# data types to store answers to questions and values of system variables. Now, those data types have simplified. The number of types have been reduced. The types now more closely match user intuition. Most users will not notice the change. Advanced users should consult this summary table of data types.
Previously, testing with Tester could be tedious. When a problem was found in a questionnaire, one made a correction in Designer, reloaded the questionnaire in Tester, and then re-entered testing data.
Now, when the user reloads a questionnaire, Tester tries to keep all previously entered data intact. That way, one can more quickly confirm that identified problems have been resolved.
Application Programming Interface (API)
In version 5.21, Survey Solutions' API have been considerably improved and expanded. First, virtually every user action now has a corresponding API (e.g., creating assignments, assigning interviews, downloading data, and approving and rejecting interviews). Second, each server features up-to-date and interactive documentation on what API actions are allowed, how to formulate them, and what server responses to expect.
Please note that this feature is aimed at advanced users who are comfortable with server queries and REST APIs.
Area question type
Working in partnership with ESRI, the Survey Solutions team has integrated some of ESRI's mobile GIS tools into Survey Solutions. The result: a new question type that measures area by tracing a shape's outlines on satellite imagery (e.g., bounds of a farmer's field). The appeal: the ability to measure area without traversing a shape (e.g., walking around a farmer's potentially distant field).
In version 5.21, this new question type should be considered experimental. In future versions, this question type will be improved and extended. Stay tuned.