(Feedback from Mexico)

Completeness of the Model

GSIM is the information infrastructure to support GSBPM.  There is a need of a mapping of every subprocess of GSBPM vs the GSIM objects. If this mapping cannot be done there should be at least one at the phase level.  This is, mapping the input and output objects for every GSBPM Phase.


Description of objects

GSBPM is a process-oriented model.  Every phase or every subprocess has a data set as input and a data set as an output.  We see two ways of incorporating the GSBPM needs:

1) General Form.  In this context we should have the following objects:

  • Process
  • Phase
  • Subprocess
  • Input Data Set
  • Output Data Set

2) Specific form at phase level:

    • Raw needs
  • Specify Needs Phase
    • Structured needs
    • Business case
    • Design Phase
      • Conceptual model
      • Logical model (business rules)
      • Statistical design
      • Sample design
      • Collect design
      • Process design
      • Analyze design
      • Disseminate design
    • Build Phase
      • Software image
      • Test report
      • Training Report
    • Collect Phase
      • Sample set
      • Collected data set
      • Collection Report
    • Process Phase
      • Processed data set
    • Analyze Phase
      • Information Set
      • Publishable data set
      • Indicators/Technical analysis
      • Aggregates
    • Disseminate Phase
      • Product
      • Presentation
    • Evaluate Phase
      • Recommendations

Option 1 is very simple but is very difficult to explain to the final user.  Moreover, it opens the door to random specification of objects. 

Option 2 is more detailed, but it is easy for a business person to understand.  We are using a version similar to Option 2.  We achieved institution-wide acceptance.

We advocate for a detailed specification of objects as the one shown in Option 2.


Consistence with GSBPM

We incline for an in-depth review in which all phases of GSBPM have an explicit relationship with the GSIM objects.  In such a review some GSIM objects might be added, in particular those which would be needed to represent the inputs and outputs from the sub-processes.

Examples of some objects that might be included are:

  • Collection instrument (as a specialization of exchange channel)
  • Quality metric
  • Questions library
  • Technical documentation
  • Validation result
  • Sample
  • Training program
  • Row data
  • Integrated data
  • Collected data
  • Integrated data
  • Validated data
  • Processed data
  • Published data
  • Disseminated information
  • Statistical indicator
  • Statistical aggregate
  • Statistical tool
  • Confidentiality rule
  • Product component
  • Statistical error
  • Evaluation report

Management of the Overarching processes of GSBPM

Information objects related to GSBPM overarching process with a statistical component could be described.  For instance:

  • Quality management case
  • Quality assessment
  • Quality control
  • Quality indicator
  • Quality principle evidence
  • Metadata custodian
  • Metadata owner
  • Structural metadata
  • Referential metadata
  • Data management process
  • Data custodian
  • Data owner
  • Data inconsistency

GSIM puts the metadata management, data management, methodological research, and design functions under the Statistical Support Program. The concept of Statistical Support Program should be deleted.  The concepts of all overarching processes should be added.


Glossary of non-information objects

There is a need for a Glossary that includes concepts that are used in more than one HLG MOS model.

The description of GSIM objects is the best Glossary we have (a very good one).  However, if we are only staying in the context of GSIM, the might be a good idea to include definitions that are not information objects. They are important to unify the criteria when describing the statistical process, for instance:

  • Reference metadata
  • Statistical framework
  • Structural metadata
  • Process metadata
  • Quality metadata
  • Statistical domain
  • Data cleaning
  • Statistical quality control

 

  • No labels