Download presentation
Presentation is loading. Please wait.
Published byJessica Porter Modified over 8 years ago
1
ROMA 23 GIUGNO 2016 MODERNISATION LAB - FOCUSSING ON MODERNISATION STRATEGIES IN EUROPE: SOME NSIs’ EXPERIENCES Giulio Barcaroli – Lessons learnt and conclusion remarks Modernisation Lab – Focussing on Modernisation Strategies in Europe: some NSIs’ experiences Lessons learnt and conclusion remarks Giulio Barcaroli| Italian National Institute of Statistics - Istat
2
ROMA 23 GIUGNO 2016 MODERNISATION LAB - FOCUSSING ON MODERNISATION STRATEGIES IN EUROPE: SOME NSIs’ EXPERIENCES Giulio Barcaroli – Lessons learnt and conclusion remarks Standardisation of internal production processes: no more pipelines with individual solutions, but an integrated and industrialised system Paradigm shift in data collection: use those already available before asking for costly and burdensome new ones Paradigm shift in competencies: from a methodological expertise able to handle the classic survey production process, to the new capabilities required to build a registers based production process Paradigm shift in methodology: from the traditional (design based / model assisted) estimates produced by sampling surveys, use of model based estimation systems applicable in a multi-sources environment 2 Different (but complementary) concepts in this session presentations What is “modernisation” in official statistics?
3
ROMA 23 GIUGNO 2016 MODERNISATION LAB - FOCUSSING ON MODERNISATION STRATEGIES IN EUROPE: SOME NSIs’ EXPERIENCES Giulio Barcaroli – Lessons learnt and conclusion remarks Official statistics production processes are often characterised by non optimality in terms of both effectiveness and efficiency Standardisation can greatly help to increase optimality, whenever optimal methods and instruments are adopted as standards, and processes are designed following a well-defined Business Architecture Yes, BUT: standardisation decreases the degrees of freedom in a system A certain degree of freedom should be granted to explore new solutions National Statistical Institutes are not only producers, but also researchers: creativity is vital 3 One major issue: standardise, yes, but to what extent? Standardisation: yes, but …
4
ROMA 23 GIUGNO 2016 MODERNISATION LAB - FOCUSSING ON MODERNISATION STRATEGIES IN EUROPE: SOME NSIs’ EXPERIENCES Giulio Barcaroli – Lessons learnt and conclusion remarks Surveys are characterised by high costs and decreasing response rates Increasing availability of administrative data and new data sources (Big Data), less costly and not burdensome Yes, BUT: this sources are characterised by the fact that they are not under the control of statisticians Problems: representativeness, different definitions and classifications, always-changing mechanisms of generation 4 They seem to refer to a common view: the classic (sampling) survey is out of date. In some cases, to be abandoned. Paradigm shifts: no more surveys?
5
ROMA 23 GIUGNO 2016 MODERNISATION LAB - FOCUSSING ON MODERNISATION STRATEGIES IN EUROPE: SOME NSIs’ EXPERIENCES Giulio Barcaroli – Lessons learnt and conclusion remarks We need a bridge between what is under control and well known (the survey process and the survey data) and what is neither one nor the other Survey data to build models… … to be applied to data from other sources… … to produce more statistical information, less costly, more timely and possibly of higher quality 5 The way to go: a combination of sources and processes, in which the (sampling) survey plays a fundamental role A combination of survey data and new sources
6
ROMA 23 GIUGNO 2016 MODERNISATION LAB - FOCUSSING ON MODERNISATION STRATEGIES IN EUROPE: SOME NSIs’ EXPERIENCES Giulio Barcaroli – Lessons learnt and conclusion remarks In any case, survey process and survey data will be only a component of the whole system New sources and different processes ask for new competencies Not only sampling methods and statistical modelling, but also: machine learning (statistical learning), text mining, network analysis, … 6 Not only statisticians, not only IT experts: data scientists Data science? Yes, without “but”
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.