Development and Structure of the Memobust Handbook
Leon WILLENBORG, Sander SCHOLTUS, Arnout VAN DELDEN[1]
Division of Process Development, IT, and Methodology, Statistics Netherlands, P.O. Box 24500, 2490 HA The Hague ()
Summary: The main goal of the Memobust (Methodology of Modern Business Statistics) project is to update an existing handbook on the methodology and design of business statistics [5]. The new handbook is actually written from scratch, and in a new form. The content is structured according to the GSBPM. The Memobust handbook is not a traditional handbook as the original one is. It currently consist of a large collection of separate modules in the form of PDF files, made available on the internet. Each module is about a particular topic. This modular approach should facilitate easy maintenance of the handbook. It also is a first step to the intended incarnation of the handbook, namely as a wiki. In this form the handbook should be easily maintainable. In the talk an overview will be given of the topics covered by the handbook. To assist the writing of the modules two templates have been designed, one for themes and the other for methods. These will be briefly discussed. The Memobust handbook is not intended to be a closed and static object. On the contrary, it is conceived as a living thing, which needs constant updating and maintenance in order to keep up with new insights, developments and technical possibilities in its focus area. This requires active input from the target group, especially in the period following the completion of the Memobust project (December 2013). In particular other ESSnet projects are invited to contribute to the handbook. Other contributions, however, are valued as well.
Key words: methodology, business statistics, design, data processing, knowledge representation.
1. Introduction
This paper describes the Memobust project: what it is supposed to produce and the way it tries to achieve this. The main goal of the project is to update an existing handbook [5] on the design and methodology of business statistics.
The paper is organized as follows. In section 2 the Memobust project is briefly discussed. In particular the aims of the project are explained and the way in which we try to achieve them. In section 3 the handbook itself is the central topic. It is explained which topics the handbook is supposed to cover, for which target group it is intended, what its desired appearance should be, how it is written and reviewed, etc. Section 4 discusses the role that the Memobust handbook can have in other activities. In the final section, section 5, we try to look into the future and indicate how we wish that the handbook will develop after the end of the project, in particular concerning its contents.
2. Memobust project
2.1 Project information
The Memobust project is an ESSnet project, in which the national statistical institutes (NSIs) of 8 countries participate: Norway, Sweden, Poland, Hungary, Greece, Italy, Switzerland, and the Netherlands, which coordinates the project. The project started formally at the end of December 2010 but de facto in January 2011. The first phase of the project will end in June 2012. The second and final phase of the project will end in December 2013.
2.2 Main project goals
The main objectives for this project are the identification of best practices and the development of common methodology and ESS guidelines supporting the production of business statistics aiming at reducing respondent burden and fostering efficiency and integration of processes. These objectives are made operational by the project goal of updating an existing handbook on business statistics [5]. After about 15 years, this handbook has become out of sync with statistical practice, due to many developments that have taken place since its completion. [5] is a handbook in the traditional sense of the word, which means that it is a monolithic structure. The Memobust project aims to update this handbook to the current practices in the area of business statistics. In fact, ‘update’ is somewhat of an understatement, as the structure of the new handbook will be very different from the previous one; see Section 3.1.
As the handbook is intended to be used in principle by countries in the European Statistical System (ESS), it is important that it should find an important base in this community. The challenge here is to balance the interests of the intended readership, which is a rather broadly defined group; more on this in Section 3.3. The Memobust handbook is currently maintained at a website hosted by Eurostat; see Section 3.7 for more details.
3. Memobust handbook
3.1 Form and appearance
The Memobust handbook in its intended form is not a traditional book like its precursor [5] and other existing handbooks on business statistics, such as [1]. Instead of being a monolithic structure, it consists of a set of separate, but interconnected documents. Most of these documents are so-called modules. In addition, there are a few documents that serve as introductory, contextual or background material.
Within the first phase of the Memobust project the handbook will be realized as a collection of separate PDF files, put on a web site. The idea is to use this material to produce a wiki in phase 2 (see Section 3.6).
3.2 Templates and modules
There are two types of modules in the Memobust handbook: themes and methods. Roughly speaking, themes are less specific and more verbal pieces, that aim to discuss a common point in a general, non-technical way. They point out, for instance, what certain techniques have in common, why they are used, etc. Methods are more specific, and usually more technical in nature. Themes should be suited for a rather broad readership, whereas method modules are more intended for specialists, such as methodologists. Both types of contributions are standardized in form. They are written using templates, which have been especially designed for the handbook.
Through the use of templates we try to structure the modules. This ensures that certain information is provided about all methods and themes, and that this information can be found at fixed places within the modules. It does not mean that a single author has to provide all the information required. Information can be added or modified later by other experts. Another advantage of this structured approach is that the modules can become accessible in various, alternative ways.
3.3 Intended readership
The readership of the Memobust handbook is intended to be rather broad. It is primarily aimed at persons working at statistical institutes in the area of business statistics, in particular survey managers and statisticians involved in the production process. Moreover, the handbook should especially be useful for methodologists.
The initial idea to serve such a broad spectrum of readers was to make all modules non-technical, or at a level requiring modest and basic technical knowledge. However, while the handbook was being written this plan seemed to fail, as several topics were actually rather technical and required the use of formulas and a decent mathematical sophistication. It turned out that the theme modules are not technical, and of a rather general nature. They should appeal to the entire intended readership, and in particular to the statisticians and the managers. Most method modules should appeal mainly to methodologists, as the name suggests. For that reason they are allowed to be a bit more technical. However, they still should provide an overview and a guide to the relevant literature where the details can be found.
3.4 Constraints
The goal is to make the terminology in the handbook consistent/compliant with the Generic Statistical Business Process Model (GSBPM – see [4]) and the terminology of SDMX (see [2]). This is more easily said than done. It requires an extra effort to obtain this consistency. In phase 2 attention should be devoted to this aspect. It should be an explicitly defined task to check (all or some) modules, assess the quality of the terminology used (with respect to compliance to the standards mentioned), and come up with a plan to improve a module if necessary, and/or actions to improve the standards.
3.5 Editorial board and the internal review process
The Memobust handbook is developed by realizing a number of modules. Which modules will be written, is specified in a work plan. An internal review process has been set up to create some form of quality control for the handbook. All modules that have been written have to be reviewed at least once, and ideally at least by two persons.
An editorial board has been created in phase 1 of the project. It has a task in case a conflict arises between an author and a reviewer (e.g. an author refuses to implement a recommendation of a reviewer), or in case the editorial board finds that an author’s work is not quite up to standard. The members of the editorial board also have a role in integrating the various modules into a more and better connected unity.
3.6 From modules to a wiki
At some point in phase 2 of the project the written modules will be transformed into a wiki. One in a wiki form, efforts can be developed to increase the coherence of the material. From then on new contributions will be done directly in the wiki.
It has to be decided yet where the wiki will be located. Eurostat is a possibility. But then it has to be clear that their wiki is suitable for the contents of the modules. It should be able to adequately present formulas, figures and tables.
3.7 Website
The modules of the Memobust handbook that have been written and possibly internally reviewed, are put on the project’s website, which is part of a larger internet portal concerning all ESSnet projects. The website can be found at http://www.essnet-portal.eu/memobust-0. Apart from the modules of the handbook, other project information can be found there as well. Some of the information on the website is public, and some is private, i.e. only visible to project members and/or members of other ESSnet projects.
Currently, the modules that have been written so far are only visible to project members, because many of them are still in an early stage of development (e.g. not yet reviewed internally). After the completion of phase 1, a first, partial version of the handbook will be put in the public section of the Memobust website. This public version of the handbook will be updated and supplemented with new modules as phase 2 of the project progresses.
3.8 Contents
The content of the handbook is roughly structured according to the components of GSBPM. Table 1 lists some of the topics in the handbook to give an idea of the content as it is foreseen so far. Most topics are subdivided into several modules. Please note that this table is not an exhaustive list. A complete overview of the planned contents of the handbook can be found on the project’s website.
4. Possible use of the handbook
In this chapter we consider some potential uses of the Memobust handbook. We start with the obvious one: the handbook as a source of knowledge to assist those working at statistical offices in the area of business statistics.
Table 1. Non-exhaustive overview of the contents of the Memobust handbook
Topic / DescriptionUser needs / Exploration of user needs; specification of intended output
Overall design / Optimisation of survey design under restrictions, including allocation and balancing for sub-processes
Repeated surveys / Design, running, and improving of surveys that are run repeatedly, e.g. for measuring changes over time
Statistical registers and frames / Description of the role of frames, unit types, building and maintaining statistical registers for business surveys; coordination across frames
Collection of primary and secondary data / Design of data collection with attention to mixed mode designs; data collection instruments (CATI, CAPI); strategies for using secondary data
Micro and macro integration / Micro integration methods (linkage, statistical matching of data sources, reconciliation); methods for integration at an aggregated level (Stone, Denton, and others)
Data editing and imputation / Overview of data editing methods for primary and secondary data; model-based and donor-based imputation methods; imputation under restrictions
Weighting and estimation / Weighting methods to obtain estimates from a sample survey; design-based, model-assisted, and model-based estimation, including small-area estimation
Quality assessment / Assessment of product quality such as variance estimation, assessing non-sampling errors, analysis of revisions, timeliness and coherency
Seasonal adjustment / Main principles of seasonal adjustment methods; treatment of time-series models, back- and forecasting methods, outlier treatment
Other topics / Questionnaire design; Design of statistical concepts; Quality reports and frameworks; Response; Sample selection; Coding; Statistical disclosure control; …
4.1 Aid in designing and running business statistics
This is the basic purpose of the handbook. It should be useful for those working in the area of business statistics. This is where we position our main target group: employees working at a (national or supra-national) statistical institute, such as methodologists, statisticians, or managers. This group includes professionals who already have some experience working in this area and want to increase their knowledge of certain topics.
4.2 …Or when learning to do so
A special group within statistical offices consists of novices, who are new to official statistics – or at least to business statistics – and want to have a quick overview of the area, before delving into one or more topics. The handbook should provide them with an overview of all topics that are important at the moment, initial information about these topics and pointers to the literature for those who are interested to delve more deeply into a particular topic.