JSM 2024: Presenting a confidence interval tabulator for differentially private microdata

We’re super excited to exhibit for a third year at the American Statistical Association’s Annual Joint Statistical Meetings in Portland, Oregon, August 4th-8th, 2024.

Cantabular is high performance, multi-lingual data publication software that offers real-time privacy protections to speed up data dissemination and to provide safe researcher access to confidential data.

Preview: a confidence interval tabulator for differentially private US Census data

At JSM we’ll offer visitors an early preview of a confidence interval tabulator built with Cantabular for differentially private US Census data. It provides non-expert users access to confidence intervals for the privacy protected microdata file and approximate Monte Carlo files.

• Supports non-expert users

• Provides a simple interface for users to build their own tables for any geographic level

• Calculates low and high confidence intervals for each official statistic

• Exactly replicates bias-adjusted Wald confidence intervals

• Offers pivoting and filtering

• Allows selection of geography using a map

• Super fast downloading for use elsewhere

Live dissemination: for ONS, UK and NISRA

We will showcase live systems powered by Cantabular for the protection and dissemination of Census 2021 data for England and Wales (ONS, UK) and for the Northern Ireland Statistics Research Agency (NISRA).

• Office for National Statistics, UK Create a custom data set is powered by Cantabular and applies disclosure avoidance in real time using our Disclosure Rules Language (DRL).

• Northern Ireland Statistics & Research Agency (NISRA) Flexible Table Builder built on Cantabular is a repeatable analytical pipeline for Census 2021 data with disclosure avoidance applied in real time.

• Northern Ireland Census 2021 mapping application pulls data directly from NISRA’s live dissemination system and was built in two weeks using Cantabular’s APIs.

Automate output checks for researchers

Cantabular helps users produce safe outputs fast and with minimal intervention. It works with many types of data: survey data, magnitude data, synthetic data and differentially-private data.