GSBPM: Other feedback

[Chris-ECE]

Suggestion to rename GSBPM

ONS
Is GSBPM a misnomer? Does it actually cover business processes? It is often still referred to as statistical value chain in some places which seemed more motivating and memorable Rename GSBPM

Comments about merging GSBPM and GAMSO

Romania
The stage of Implementation of GSBPM and GAMSO at the level of the National Institute of Statistics Romania Integration with GAMSO
The implementation of the two standards at the National Institute of Statistics level represented a challenge in view of the complexity of the activities carried out at the level of the National Institute of Statistics. In the production process of statistical products were identified both activities of production of statistical data and support activities (legal, IT, accounting, human resources, etc.), without which the institution would not be able to carry out its activity. Integrate GAMSO & GSBPM
Therefore, in order to implement the two standards, the decision was made to implement an integrated vision. Thus, for each sub-phase associated with both GAMSO and GSBPM, specific generic activities were developed and defined. Integrate GAMSO & GSBPM
This list of activities is structured in 3 parts: GAMSO/GSBPM phases, GAMSO/GSBPM sub-phases and specific generic activities associated to each sub-phase. Integrate GAMSO & GSBPM
The phases associated with the two standards are coded by alpha-numeric codes (A-C- activities specific to the GAMSO standard, D-K- activities specific to GSBPM). This list of activities is called the "List of Generic Activities" (LGA). Numbering Integrate GAMSO & GSBPM
The implementation of the two standards aims at two extremely important objectives:

1. Development of an IT application for calculating the cost of surveys and statistical research;
2. Standardization of statistical methods used in the data production process. Can we look at a proposal?

ONS
Should GSBPM and GAMSO be integrated into a single activity model? Integration with GAMSO

Mexico
• Considering GAMSO and GSBPM as complementary models, it seems convenient to align the relationship between their respective concepts and components. It is necessary to establish a better balance between both models and to standardize the language that is used in both. A schema of the hierarchies of Phases, Subprocesses, Activities, Tasks, relating both models would be helpful for contextualizing and interpreting them. GSBPM-GAMSO unified schema
• An emphasis on how GAMSO provides the organizational framework to perform the GSBPM would help to understand the value of both models. It is recommended to work with both models as an integrated model defining the core components and processes of any NSO. GSBPM-GAMSO

Hungary
 We suggest integrating the activities from the GAMSO and the GSBPM to create one generic activity model (example* attached: 1. activity model) that covers an entire NSO. GSBPM is already part of GAMSO (as far as we know, the original intention was that after the development of GAMSO, there might not be two separate models). We suggest the purpose and the relationship of the models to be (re)defined: GSBPM as a statistical process model and GAMSO as a collection of activities grouped by activity area.) Models should be integrated but separate???
We would find it useful to:

• match the level of detail for the activities, so the integrated view is internally consistent; General
• prepare definitions for all activities and define the most important attributes for each object, e.g.: name, definition, ID, type, etc. General
• make sure all activity areas cover the PDCA cycle; General
• include decisions, controls etc. wherever necessary; General
  Explanation: activities listed in GAMSO provide inputs for the statistical business process (methodologies, tools, HR skills, etc.) that can be attached to the process elements described in the GSBPM. In this sense, Corporate Support areas can be interpreted as individual PDCA cycles in themselves. For example, Manage Statistical Methodology describes how methods are developed (PDCA: starting from a need to develop a new method or update an existing one, and the output is a methodological instrument). Then later, when GSBPM phases and sub-processes are described, they can explicitly refer to the metadata, quality, methodology, etc. elements that are produced in a specific activity area. Integration of ModernStats Models
  The output could be used by NSOs for organizational development, for integrating new activities, restructuring the organization, etc.
  • We suggest the GSBPM should focus on how one can put together a statistical business process using elements, activities from the GAMSO and on the inputs and outputs: Integration of ModernStats Models
  • in addition to phases and sub-processes, we suggest to prepare ready modules by combination of two or more elements that are always used for certain statistical functions (such as preparing draft outputs, run dissemination processes, etc.). (example* attached: 2. process model). Linkages of subprocesses
  • also, we suggest that GSBPM should define the inputs and outputs of sub-processes (not in GSIM terminology, but in statistical context; e.g.: edit and impute uses raw microdata sets (Input = GSIM Data set) and produces a new version of the data set with edited/imputed data with relevant metadata (Output = GSIM Data set). Integration of ModernStats Models
  The output could be used by NSOs to model and/or build processes, for standardizing their processes, etc.
  General comments on the models (GSBPM and GAMSO):
   We suggest to clarify what unit the model and each item in the model (e.g. process, phase, sub-process, task, activity area, etc.) refers to. Currently there are missing definitions, the unit is not specified, and based on our experience, this causes confusion during implementation. Unit?
   Definitions/descriptions and explanatory texts: we suggest to revise the definitions/descriptions for all elements of the model(s), to balance their length/depth and to match the level of detail so the models are internally consistent. Currently some descriptions are more detailed (e.g. GAMSO “Strategy and leadership”) while others are very generic (e.g.: “Ensure general coordination and alignment” under GAMSO “Govern and lead”), also, some refer to higher level (e.g. “Conduct evaluation” in the GSBPM), whereas others refer to lower level activities in the hierarchy (e.g. “Calculate weights” in GSBPM). GAMSO consistency
  GSBPM-related comments
  We suggest:
   the model to focus on how one can put together a statistical business process using elements, activities from the GAMSO (to prepare modules by combination of two or more elements that are always used for certain statistical functions, example* attached: 2. process model) and define the inputs and outputs of sub-processes (not in GSIM terminology, but in statistical context; e.g.: edit and impute uses raw microdata sets (Input = GSIM Data set) and produces a new version of the data set with edited/imputed data with relevant metadata (Output = GSIM Data set) (see above, under General comments). Inputs and outputs Integration of ModernStats Models?

Comments that are beyond the scope of the current GSBPM revision

Statistics Portugal
• We a suggest adding schematic examples of statistical operations, to present the model’s applicability to the business process at organisations in a simpler way. One aspect is the model’s flexibility, which translates into a possibility of different sequences, iterations and combinations of phases and sub-processes. Another aspect is the model’s applicability to different kinds of statistical operations (primary and derivate statistics) and data sources (data collection through sampling, exhaustive surveys, administrative data or other sources). These aspects are mentioned along the document but would benefit from a visual help. Illustrative examples/Use cases
UNECE response: A diagram has been added to illustrate how production may iterate around GSBPM sub-processes in a non-linear way diagram. However, use case examples would be outside of the scope of this document, though they could be considered for future separate work.

• We suggest including references to the quality and performance indicators of the model at phase and sub-process level. Quality indicators
UNECE response: There existis standalone work on quality indicators for GSBPM sub-processes, released in 2012. This may be revisited by NQAF experts in 2025.

Canada

1. Record Linkage - To deal with some of the shortcoming of the GSBPM in terms of record linkage, the organization in 2017 created the Record Linkage Project Process Model (statcan.gc.ca), which was published on our StatCan website for others to use and leverage. Each of these models focusses on a different aspect (GSBPM-Surveys; RLPPM – RL), given where we are going re: use of admin, linked, big data, modelling, seems we need a model that brings all these elements together and walks individuals through the decision-making process of what approach to take when perhaps starting with admin data (if available) and use surveys as a last resort. Incorporate Record Linkage Model (RLPPM)
   UNECE response: The GSBPM is conceptual and not implementation-focused, so having a model that is both generic but able to walk people through all the decisions they need to make would be beyond the scope of GSBPM itself, although it could be something for separate future work to consider.

Feedback from Operations Perspective at Statistics Canada:
• Each task in the schedule is linked and named in accordance with phase and/or sub-process in the GSBPM. This linking process is not always straightforward because the terms used in the GSBPM are generic/vague while the operational tasks are very specific. It would be beneficial to have a GSBPM supporting document which provides examples of operational activities in every business phase. This will allow us to build common language within the organization and facilitate internal communication of operational requirements. Supporting guidance for implementation
• In order to easily identify interruptions to production (due to changes), it would be beneficial to have a supporting document which lists possible process flows. Supporting guidance for implementation
• It would be beneficial for the GSBPM to list corporate tools available in each phase and/or sub-process. This would promote the use of corporate assets and minimise the development of tools with duplicate functionality. Software tools by phase
UNECE responses: There is a list of open-source tools relevant to different parts of GSBPM (the "awesome list"). We consider that this comment might be less relevant to GSBPM than perhaps to CSPA (if a new version of CSPA or something similar is produced in the future).

• It would be beneficial to have an internal GSBPM supporting document and training that lists common Information resources of business value (IRBV) created in the GSBPM process and provides guidance on where to store them and naming conventions for their identification. Supporting guidance for implementation
• The GSBPM focuses on linear data processing. The Integrated Business Survey Processing (IBSP), which is the generic system to process business data at Statistics Canada, is not a linear process, it is predominantly focused on iterative processes that combine continual collection, processing, analysis, and partial dissemination phases together.
UNECE response: A diagram has been added to illustrate how production may iterate around GSBPM sub-processes in a non-linear way diagram. This should make the non-linear nature of GSBPM more apparent.

France
• A document that lists the different implementation cases of the GSBPM would be very useful; Use cases
• If processing, it could be useful to clarify interactions between the sub-processes of 4 and 5.
20. Statistics Ecuador (received: March 16)
5. Disseminate phase
In this phase, a sub-index could be added describing the minimum products required for its implementation. On the other hand and regarding the updating of products related to dynamic and interactive charts, it is suggested to mention what software is available to use for this purpose.
In general, the document could be strengthened including the next scheme for each phase: Inputs and Outputs

This scheme enables the producers of statistical information to understand clearly the documentation that could be generated from the entire production process, and the activities involved in each process. The main idea is to generate, from theory, an empirical document, easy to understand, a guide regarding the products and activities resulting from each phase and sub-process. Phase 7 7

Comments that are not directly actionable

Sweden
At Statistics Sweden, we have just finished our work with updating our process model to support a modernised production process where we work with a mix of different types of data sources, with a clear strategy of re-using existing data first, then using registers and administrative data and direct collection as a last resort for data capture. Furthermore, we apply a more flexible approach to statistical programs where different programs will have somewhat different tasks; some programs will capture data and produce intermediate registers of data, registers that other statistical programs will use to produce final observation registers and statistics. With these main considerations, our updated process model differs a little from the existing GSBPM. The major differences (there are also minor differences in the naming and content of sub-processes, and we have reduced the number of sub-processes) are the following: non-survey data source, flexible Intro text
An overview of our new statistical production process at Statistics Sweden can be found below.

Mexico

• INEGI has approved “guidelines for surveys” which detail tasks corresponding to sub-processes 2.1 Design outputs, 2.2 Design variable descriptions, and 2.4. Design frame and sample. We believe that these guidelines could be a useful input for this GSBPM revision.
• The work that has been made to relate GSIM and GSBPM can provide valuable information about typical descriptions of each subprocess, their inputs and outputs. We believe that the Linking GSBPM and GSIM report 2021 could be a useful input for this revision. Inputs and Outputs

Hungary
 Regarding the descriptive parts of the document (Chapters I., II. III., VI and VII.), we support keeping the history, the connection of the model to other ModernStats models and highlighting the most important changes to the previous version. We suggest the other chapters to be revised.
 Emphasizing the statistical perspective of the models. We suggest the revised model(s) should be specific for statistical organizations as much as possible. In case they become too generic, statistical organizations “will not recognize themselves in the story”.

Italy
It would be great if this revision of GSBPM and GAMSO could solve some terminology issues that are incoherent in Modernstat Models.
UNECE response: Various changes to terminology have been made, which hopefully address this comment.

Portugal
• We suggest updating the content concerning the SDMX in the point 125 by mentioning the new 3.0 version and highlighting the expansion of the technical specifications for geospatial metadata that includes geographic features of statistical data and enables to link statistical metadata with geospatial metadata.
UNECE response: That line no longer exists (the latest version of SDMX is 3.0, but there is no SDMX 3.0 user guide).

Suggestion for a future version of GSBPM, based around "Steady States"

Norway
Steady states of data are organized according to GSBPM in the new data platform that we are currently building. In the longer term, it may be appropriate to update GSBPM with a more concrete description of "new data sources”. non-survey data source
There should be good statistical reasons for making changes to the model. It has been adopted by many statistical organizations that Statistics Norway collaborates with in Norway and other countries, including in aid projects.
UNECE response: We agree it may be useful in future to consider this.