4 - Understanding Attribute Disclosure

Attribute disclosure is when the data allows for an inference of sensitive information about an individual without identifying the individual in the published data set.

For example, you are working on a project requiring the execution of analysis against a De-Identified Data set. As part of the analysis, you must append additional fields to the De-Identified Data set using the Indirect Identifiers to match records. This is an example of Attribute Disclosure Risk.

Attribute disclosure occurs when newly discovered characteristics can be determined about an individual. The likelihood of Attribute Disclosure is determined based on how easy it would be to use the information in the data set to combine it with other information. 

An example is Identity Disclosure - linking a de-identified respondent to a dataset containing the individual’s identity.

Did you know that it is against our contractual language for EA staff to try and re-identify the data used in our products AND it is also against our license agreement for our clients to try and re-identify EA's data products?

Attribute Disclosure and EA

De-identification techniques protect against the disclosure of individuals’ identities and linking information to them. They do not, however, protect against the disclosure of attributes relating to groups of individuals that may be stigmatizing.

As a best practice, we will consider whether any group attributes in a de-identified data set are stigmatizing before releasing the data set.

How do we keep ourselves from inappropriately using our data?

Your license agreements address the risk of the use of EA data, where an over-representation of certain attributes, such as age, gender, and income, used in an inappropriate way may perpetuate neighbourhood stigma, gentrification, redlining, and other systemic discriminatory harms.

Congratulations! Now you understand the basics of Ethical Data Use and how it applies to using EA data.

Your training objectives were for you to be able to:

  1. Understand Potential Stigmatization & how it applies here at EA.
  2. Understand Attribute Disclosure & how it applies here at EA.