Page 55 - Invited Paper Session (IPS) - Volume 1

IPS57 Eric Rancourt

Sound Methodology for Sound Official Statistics
Eric Rancourt
Statistic Canada

Abstract
The information society has evolved to a point where data and pieces of
information of all kinds are readily available at our finger tips. All these digital
fragments can be informative or misleading. It is simply not possible to
understand and/or defend a data point when it stands on its own; it requires
a context and a purpose to become a meaningful piece of information. In this
data day and age, contexts, purposes as well as tools, technology and methods
are overwhelmingly abundant. However, how does one know which are the
most appropriate for their purpose? While National Statistical Offices (NSO)
might have been lighthouses whose beacon people readily followed, they now
have to explain and justify themselves for constituents to see the value of the
information they emit. It is the responsibility of the NSO to be able to not only
produce quality data, but also to enable potential users to understand what it
produces. This article will use the example and values of the Methodology
Branch to provide an example of a frame for meeting some of the rigour and
transparency challenges that NSOs are facing.

Keywords
Valid statistical inference; Rigour; Quality; Scientific approach.

1. Introduction
National Statistical Offices produce information that constitute Official
Statistics. This information stems from data needs that are usually expressed
in statistics acts (e.g. Canada, 2018) or from data needs expressed by decision
makers for which data was gathered and then statistics produced. The
description of processes and methods used to produce such information is
usually provided and explanations are readily available to users who are
interested in increasing their understanding related to these processes (see for
example the Statistics Canada’s Policy on informing users of data quality and
methodology, Statistics Canada, 2000). Sometimes the quality of statistics or
the actual production of statistics might be questioned, requiring that the
processes and methods are not only explained/detailed but also justified and
defended. In other situations, nothing is asked and nothing needs to be
defended; it is simply that attention is drawn elsewhere - to other data sources
that may not have the quality that is produced by the NSO. In all three cases,

44 | I S I W S C 2 0 1 9