PERTS has designed reports so that data remain stable and accurate, even when a Class misses a survey. There are differences between Class & Community/Network reports, read on below.
In the Class report, if the Class missed the survey and no responses were recorded, the report will simply not show data for the survey that was missed. Charts that show change over time from one survey to the next will skip that survey entirely. For example, if all participants in a given Class take the survey for the first time at the start of Survey 2, the report will show the Survey 2 data as the baseline data.
In Community/Network reports, however, we cannot simply hide the missed survey because other Classes in the same Community or Network may have data for that survey. In that case, we use our standard methods for handling missing data to fill in values for the Class’ missing survey. Imagine a Community of Classes A-D, with surveys scheduled in January, March, and May. Class A gets a late start and only participates beginning in March. There are two ways of handling this scenario, each with similar results in a Community report:
Option 1: Class A Skips Survey 1
Class A simply starts at Survey 2, and continues implementation along with other Classes.
In this case, Class A would have their data imputed backwards, or copied over from Survey 2 into Survey 1. This stabilizes the data and ensures reports only reflect changes that have actually been observed.
Option 2: Class A Adjusts Their Survey Schedule
Another option is to change the survey schedule for Class A so that Survey 1 is scheduled for March and Survey 3 is deleted.
In this case, data would be imputed forwards instead of backwards. Community/Network reports generated at the end of March would copy Class A’s Survey 1 data into Survey 2, instead of the reverse. Once Class A completes Survey 2 in May, the imputed data would be replaced by newly-collected data. Those new data from Survey 2 would also be carried forwards to Survey 3. In this way, we ensure that every data point in a Community/Network always has a stable representation of all participants in the Community/Network and only reflect observed changes in participants’ experiences.