The innovations inventory in its current form is designed to be an ideas hub for improving official statistics. Therefore it provides short introductions to each innovation and contact details, if available, of the lead innovator.

The innovations inventory is open to everybody. There is no focus on specific regions or specific types of innovations. As long as they offer innovative solutions for official statistics, they are included. Innovations so far have been collected via a variety of channels. First, PARIS21 collected innovative approaches during the work on the ‘Informing a Data Revolution’ project. Second, in late 2014 the ‘Global Call for Innovations’ created further input for the inventory. Third, in early 2015, a user platform has been introduced to give innovators a chance to promote their ideas and projects by adding innovations to the inventory themselves. The innovations inventory is growing further every day.

As existing classifications for innovations have proven to be too general or too specific to a different topic it has been necessary to develop a new way of classifying innovations in official statistics. The classification scheme of the innovations inventory incorporates so far three distinct perspectives on innovations in official statistics. Someone who is interested in a certain field of innovations is able to search for that by type. Someone who is interested in which innovations are available in a certain country can search for that by region and producers of official statistics are able to see which innovative solutions exist for a specific statistical production process using the Generic Statistical Business Process Model (GSBPM) as search framework. While the scope of countries and the GSBPM are predefined, the typology of the innovations, however, is derived from the diversity of included innovations and thus is subject to possible future adaption.

The questionnaire is structured into five modules with 13 different themes and comprises of in total 81 questions, trying to cover the most important features of a national statistical system. Both the methodology and the questions are partly derived from ‘Snapshot’, a comprehensive assessment tool of EUROSTAT developed to assess the strengths and weaknesses of the statistical systems of EU member states. Further questions were added to strengthen both the user and the forward looking perspective and to determine future needs and challenges.

The selection of countries for the country studies is based on a stratified sample by countries' statistical capacity index and geographic regions. The statistical capacity index is based on the average capacity score for each of the six dimensions in the PARIS21 metabase. Within each strata, countries were selected at random and non-respondents were replaced with a random draw from their respective strata. In detail, the steps are the following:

1. The data used for the selection process is retrieved from the PARIS21 metabase.

2. In order to form a rating out of the PARIS21 metabase data, the data is decoded to binary, if needed, by formulating questions, e.g.: Was there a population census in the last 10 years? The binary data is then combined to the dimensions listed in the PARIS21 metabase using a weighted average. Together, the dimensions form the capacity index used for the selection.

3. Information on population, geographic region and income in form of gross national income (GNI) per capita are used to study the distribution of the capacity index. For twelve countries GNI information are not available. Hence, they are excluded from further analysis, reducing the number of countries from 136 to 124.

4. The sample design is based on the capacity index and the geographic region. The target number of countries covered in the country studies is 25. The sample count is proportional to the number of countries in a region. For example, having 17 African countries with a capacity index between 9 and 10 of a total of 124 countries equates to 3.4 (rounded 3) African countries with a capacity index between 9 and 10 in the target sample of 25 countries. Deviations might occur due to rounding.

5. Among the sample countries, the countries are selected randomly. Participation in the country studies has been offered to the respective NSOs while respecting their sovereignty. In case of non-response, countries are re-drawn. The final sample used for analysis consists of 29 countries.

6. The country selection for the in-depth country studies was conducted as described in 4. and 5. with a target sample of 7 countries and requiring participation in the cross-country study.

The IDR country studies were carried out in three stages between May and September 2014. The stages differ in depth and in scope. In Stage 1, a comprehensive database was compiled to include information on the structure, activities and outputs of statistical systems in all developing countries. Once this information was compiled, national statistical systems were divided into a small number of distinct groups with similar characteristics. Stage 2 involved a desk-study assessment including a questionnaire of a sample of countries from each of the groups, involving 29 countries in total (for details see ‘Country studies selection process’). Stage 3 involved in-depth investigation and analysis in seven countries, selected from each of the groups identified in Stage 1.

All indicators used in the PARIS21 metabase are compiled from publically available data. Links to the sources are provided for each indicator. The only exception builds the Use of Statistics indicator, which is a PARIS21 internal indicator. Data and detailed methodology is, however, available upon request. The choice of indicators is inspired by the World Bank’s statistical capacity indicator and extended after thorough internal and external evaluation to strengthen both the user and the forward looking perspective of the PARIS21 metabase.

The concept of dimensions clusters single indicators describing the quality of national statistical systems into broader thematic blocks. It is derived from the Statistical Capacity Indicator framework of the World Bank and augmented with the dimensions Use/Demand and Innovations. Other concepts for assessing statistical quality (e.g. UN NQAF, IMF DQAF, EUROSTAT QAF, Statistics Canada Quality Assurance Framework) proved less adequate for the purpose of this project, since features such as Accuracy or Consistency were in this case not reliably quantifiable.

By mandate, PARIS21 ought to focus on developing countries. Hence, the PARIS21 metabase includes only countries that are classified as up to upper-middle income countries following the Development Assistance Committee (DAC) classification of the OECD of 2014. Further, only countries with full UN membership status that are not OECD countries are monitored.

The word metabase is derived from words metadata and database, combining the two most important features of the PARIS21 metabase in one name.