Academic institutions require that researchers adhere to data security procedures which protect the privacy of participants’ data. Here are some tips pertaining to documentation and metadata that we have compiled from the Government of Canada’s Panel on Research Ethics (Government of Canada, 2016), the University of Waterloo (Human Research Guidelines and Policies, 2013), and the University of Toronto (Data Security Standards for Confidential Data in Research, 2019).
Direct identifier data sets and identity-only data sets must be stored in a secure location and in secure data-encrypted form at all times.
It is not possible to completely de-identify all research data sets (e.g., an audio recorded interview in which a participant identifies him or herself). The original data set must now be treated as an identified data set.
Collect the bare minimum of identifying information required for the study’s conduct.
Describe specifically what personally identifiable data elements/variables would be gathered, and why they are needed for the proposed study. Describe if the data set will be de-identified and identity-only, or anonymized with no identity-only data set in the application.
Even More Tips on Data Security Procedures…
Level of Security is Proportional to Type of Data
Level of security required to protect personally identifiable information is proportional to the risk posed to the participant if the information is released inadvertently or as a result of wrongdoing. Sensitive personal data necessitates a high level of protection. Identified information necessitates a lower level of security because participants have explicitly consented to being identified. Collect the bare minimum of identifying information required for the study’s conduct.
Identified data is different from identifiable data. The identifiable data is the data that can potentially be used to identify a particular person.
De-Identify the Data
Data should be de-identified as quickly as practicable after collection, and identifying variables should be separated (i.e., create identity code, destroy raw data). An identity code issued by the researcher may be included in both data sets for the purpose of later integrating the identity information with other research data, and later used to link identity data pieces back to the de-identified data collection.