GSIM is method agnostic. We should not tie it to any method. Methodology should not be part of GSIM.
Maybe should have some kind of usage example? Take a method in standard use across organisations and show how existing GSIM objects could be used. Are we the right people to create this? Maybe the methodologists should do this? More use cases in the user guide? Some material from the DDI methodology group.
Are we confusing methods and methodology? Methods are embedded in the design of statistical processes. We do not need a separate object to describe this. These are concrete. There is something abstract called methodology for which are methods which are not yet concrete (eg probability sampling which is a methodology, but there are many ways implement it). Maybe this should be represented. Is including methodology as an object now an easy way out?
How important is it to have this in the next version of GSIM? Is text with a good example enough?
We need someone to provide a real example ( Dan Gillman) and then we can look at the model.
2 Comments
Jenny Linnerud
Maybe we need Data Resource, Metadata Resource and Methodology Resource as three subtypes of Information Resource.
user-8e470
Discussion 5/9:
GSIM is method agnostic. We should not tie it to any method. Methodology should not be part of GSIM.
Maybe should have some kind of usage example? Take a method in standard use across organisations and show how existing GSIM objects could be used. Are we the right people to create this? Maybe the methodologists should do this? More use cases in the user guide? Some material from the DDI methodology group.
Are we confusing methods and methodology? Methods are embedded in the design of statistical processes. We do not need a separate object to describe this. These are concrete. There is something abstract called methodology for which are methods which are not yet concrete (eg probability sampling which is a methodology, but there are many ways implement it). Maybe this should be represented. Is including methodology as an object now an easy way out?
How important is it to have this in the next version of GSIM? Is text with a good example enough?
We need someone to provide a real example ( Dan Gillman) and then we can look at the model.