(From Alice Born) Need to review definitions of terms "Aggregate variant" and "Re-grouping variant". - Should they be redefined? (GSIM Statistical Classification Model)

  • No labels

6 Comments

  1. user-8e470

    10 Jan, 2018

    Meeting 10/1: This needs some input from others. Eva Holm can provide some input in a few weeks. We could also ask Rebecca Stoks and Alistair Hamilton. Maybe Essi Kaukonen as well?

  2. user-8e470

    20 Feb, 2018

    New Zealand:

    Issue #8: Definitions of terms "Aggregate variant" vs "Re-grouping variant"

    • [From our classification experts]: The way the regrouping variant is phrased isn’t necessarily ideal but I think the sentiment and intent of what the two definitions are saying is still relevant.
    • [From RS]: This make sense to me. If you feel the need for further clarification I would probably do this via an example than more wordsmithing. 

  3. user-8e470

    20 Feb, 2018

    Estonia:

     

    Our practices differ from Canada. Our variants are not “prescriptive” and approved as standards. They are any kind of variants/ code lists derived from the actual standard which the statistics have made for specific purposes. They can be “in-house recommendations” even though they are not standards.

     

    The idea is that all the variants belonging to a certain series can be easily searched in the classification system in order to enhance reuse. In other words, our objective is to reduce the number of “non-standard” variants by making them all visible in our database and marking those ones that are “recommendations” . We had to do this as the statistical systems are using the classifications system in the production and they would “fail” if we would take their own variants “out of it”

     

    Not the only way and not the best way, but one of the ways to handle variants J

  4. user-8e470

    20 Feb, 2018

    Sweden:

    At Statistics Sweden, classification variants are not approved as standards; not even if the classification version that the variant is derived from is a standard, which is usually is.

    In the Statistics Sweden web page, classifications and classification versions are disseminated in the classification database (KDB). Classifications variants are not included I KDB, for the external user to access.  

    In the current system, MetaPlus, most classification versions have variants. MetaPlus is used for different purposes; one is to maintain the classifications and another is to document alternative groupings (used in the documentation of the statistical product). Almost any change from the classification version requires a variant; therefore, there are many variants only created for specific products. When developing the new system, the aim is to reduce the number of variants. Some variants will be re-labeled as value domains, derived from classification versions in our on-going metadata project that have the aim to create a metadata IT-system in which newly adopted rules and ongoing development of practices will be implemented.

     

    Here is an example of an extended variant:

    Swedish example: The Swedish Standard Industrial Classification (SNI) in a way could be considered an extended variant of NACE, but is seen as a reference classification instead. SNI has one level added in comparison to NACE.

     

    Example of a Regrouping variant which does not violate the structure of the base Statistical Classification

    Statistics Sweden has those as well, for instance regroupings of SPIN (Reference classification to CPA)

     

    Example of a Regrouping variant which violates the structure of the base Statistical Classification

    Sweden does not have those variants, and we cannot see that we will need this type of variants, now.

     

    Here is an example of Aggregate variant 

    Sweden has the same type of variants for age.

     Question: Where do we sort in classification variants that just have alternative codes? Such as NACE Rev. 2 with alternative codes, NUTS with alternative codes.

     

  5. user-8e470

    20 Feb, 2018

    As discussed at the last meeting of the GSIM revision group: Issue #8 Need to review definitions of terms “aggregate variant” and “re-grouping” variant.

     

    Also, need to review the definition of “variant”.

     

    At Statistics Canada, classification variants are “prescriptive” and approved as standards.

     

    Here is an example of an extended variant:

     

    NAPCS Manufacturing and Logging – where another level was added to the base classification.

     

    Here is an example of a Regrouping variant which does not violate the structure of the base Statistical Classification

    NAICS 2012 – Goods and services producing industries

     

    Here is an example of a Regrouping variant which violates the structure of the base Statistical Classification

    Variant of NAICS 2012 - Information and communication technology (ICT) sector

     

    Here is an example of Aggregate variant – it is difficult to find a true example since we do not have many linear classifications.

    Age Group

     

     

    The current definition of variant in GSIM is vague. The following paragraph does not clearly define “standard” variants. For us, standard variants are normally based on: 1. Internationally defined variants (e.g., OECD’s ICT for industry and products, SNA concepts); 2. Based on a concept (e.g., STEM occupations and fields of study); and 3. Internal requirements (e.g., North and South Canada for justice statistics).

     

    29.     It is sometimes debated whether a classification database should be descriptive or prescriptive, the idea being that a prescriptive database will contain only standard classifications, whereas a descriptive database will also contain non-standard variants. In reality, the demarcation between standard and non-standard classifications or between these and more loosely structured groupings is not very clear. It seems, therefore, that the criterion for inclusion in the database cannot be formal status only, but just as much the usefulness and commonality of the information provided. Most of the time the departures from the norm are legitimate, made to meet specific producer requirements or user needs. In any case alternative groupings exist and have to be documented. Indeed, listing the non-standard variants used in a statistical office may be a first and necessary step towards reducing their numbers.

     

    In Canada, our objective is to reduce the number of “non-standard” variants; and to maintain only “standard” variants in a prescriptive database.

     

    I am interested in your experiences and suggestions on the need to revise/clarify this section of GSIM-Classification model.

     

  6. InKyung Choi

    21 Feb, 2018

    Meeting 21 Feb.

    • Countries have different interpretation of variant and standard variant, and practices how they store/record variants
    • Examples of types of variant mentioned in current GSIM Statistical Classification Model do exist
    • Then what do we want to do in GSIM regarding variant? 
    • Variant is not an object, it is an attribute of object. Description seem okay as it is. We don't need to modify.