The Ed-Fi 2.0 data model entity AssessmentReportingMethodType contains an odd enumeration of values that suggest modifying the model.
The first issue is around including vendor-specific types in the model.
The AssessmentReportingMethodType values includes values that communicate a specific provider score (e.g. "ACT score", "College Board examination scores", "International Baccalaureate score"). There are also values that simply seem to communicate possible values (e.g. "pass/fail" or "Number score").
This model raises logical questions, such as:
- Why isn't an ACT (or other) score just a "Number score"?
- If I am a vendor and have a particular way of reporting results, do I get my own type?
- What happens if a vendor score system changes? If ACT introduces a new scoring system in 2018, is there a new type called "ACT score 2018"?
It seems like the field/type exists to help downstream applications "interpret" the assessment results provided. If that is really the case, then communicating that a value is an "ACT" value or "College Board" value may best be done in some other fashion. (For example, a "provider" and a year may help to make the connection in a more scalable way).
As a secondary issue, a number of the type values suggest that the field is trying enumerate reporting types to provide value to results interpretation. But conceptually, many type/enumeration values do not actually contain enough data to interpret results, or - alternatively - that the number of reporting types - "percentile" "letter grade" etc. - are theoretically limitless.
To take the "percentile" example: if I was a client application receiving Ed-Fi data and I and received "percentile" results for an assessment and wanted to generate a display of how students did, I need to understand the population against which that percentile was generated - is that "percentile" for all assessment takers, for all assessment takes this year? for my state? school? etc.
Likewise, is my "letter grade" on an A, B, C, D, F scale, or a A,B,C,D scale or other?
I think the task here is to assess what projects make use of this entity, what they are trying to accomplish, and if this is a scalable way to deliver the needed data.