Data disclosure control Nordic Forum for Geography and Statistics Stockholm, 10 th September 2015
Index Main principles behind data disclosure Confidentiality in SSB Norway The board of confidentiality at Statistics Norway How to publish map data Confidentiality threshold, 2006 (Preliminary conclusion) Confidentiality threshold, 2010 Confidentiality threshold, just a number?
Main principles behind data disclosure Small population counts results in a risk for identifying individuals It is prohibited to publish official statistics, if that information can be traced back to: –a physical person in case it will lead to harm him/her. –a legal person if this will lead to unreasonable damage for him/her.
Confidentiality in SSB Norway Publication of official in statistics Norway. The guidelines are governed by the Statistics Act § 2-6. Publication of official statistics: 1. Publication of official statistics is governed by the Statistics Act § The main rule is that information shall not be made public so that it can be traced back to the respondent or other identifiable individual Marking: 9. Statistics Act § 2-6 means that the SSB in generally do not publish tables with less than three units in a group (table cell) where the sample creates the risk of identification 13. It is important to ensure that information for groups with less than three units, can not be detected indirectly by aggregates in Table Office. It is eg. no intention to tingle in the number only in one cell, if it can be read by calculating it from an aggregate table Industry statistics: 15. In industrial statistics, there are deviations from the general rule of three units in a group. In a business table, a table cell will have to contain businesses within at least three firms to sustain above safety with three units. The same is done for example. In table of commodity production in industry. Production information for a product can not be published if this is not produced in at least three undertakings 17. To complicate detection it is required that the sum of variables (eg. salary sum) in companies in an enterprise constitute a maximum of 90 percent of the total for the group. Example. If an entity has 90.2 percent of the salaries sum in an industry in a county, the figures for this group are not be published 18. Furthermore, there shell not be published figures if the two largest units (enterprises) together hold at least 95 percent of the total Trine
The board of confidentiality at Statistics Norway Mandate: Confidentiality committee is an advisory committee for the subject assessment of confidentiality when lending of micro data and publication of detailed statistics. The Committee shall: Treat and set for DM applications from external devices to achieve status as a device that may obtain micro data for research Updating a public list of devices that can receive microdata from SSB to research Give statements to the individual statistics section that processes and decides applications for loans of micro data for a specific project Consider proposals to publication of statistics that are believed to violate the general rule of the Statistics Act § 2-6 Composition: The Committee is chaired by a lawyer at the Department of Administration. All statistics departments, Division of statistical standards and methods and Communications Department is represented in the committee. Research Department participates in dealing with cases of microdata.
How to publish spatial statistics Procedure at Statistics Norway Publishing at Kart.ssb.no in accordance with “Geodataloven”: (product specifications, design standard, metadata) 1.Permission application for publishing spatial statistics What and how to publish? 2.The board of confidentiality 2. Lawyers at Statistics Norway Approved Restrictions? Rejected
Confidentiality threshold, 2006 GRIDS 1km x 1km for theme population The board of confidentiality at Statistics Norway concluded that; Preliminary conclusion For publishing of population statistics (number of persons) per 1 square kilometer, it should for the time being, not be given exact values for grid cells that contain less than 10 persons. It means that grid cells will operate with the values 0, 1-9, 10, 11, 12 and so on. For grid cells with 1-9 persons the value is set to 5.
Confidentiality threshold, 2010 The board of confidentiality at Statistics Norway believes there are grounds to publish statistics at 1kmx1km: without suppression for following themes; –Population total and distribution by sex –Housing/building(variable household must be used with caution according to board) –Agriculture with suppression for following themes; –Population average age, which requires at least 5 persons to be granted The board of confidentiality at Statistics Norway believes there are grounds to publish statistics at 250mx250m: without suppression for following themes; –Population total
Confidentiality threshold, just a number? Not for the user (Norway) Thresholds Percentage of suppressed population total population0,30 %1,30 %3,30 %5,00 % 65+2,40 %6,10 %13,50 %19,50 % Male 65+4,50 %11,60 %25,80 %38,30 % cells total population27,50 %50,00 %67,10 %73,40 % 65+60,90 %75,90 %85,90 %89,40 % Male 65+71,90 %83,90 %91,60 %94,70 % Source: University of Southampton, Statistics Norway, AIT, Eurostat
Confidentiality threshold, just a number? Nor for the suppressed population 30 minutes travelling distance from Emergency hospitals (Norway) Source: GEOSTAT 1B