" "   " "
  " "  
Home
MLV1
Debates
Motives
Selection Criteria
Managing Data
Anonymisation
Costs
Lessons

Anonymisation

 

The degree to which data need to be anonymised in relation to archiving is still a topic of debate and methods of anonymisation are in development - a process to which we are making a contribution. Making the dataset available for use by others raised important questions about how to maintain the anonymity we had guaranteed our young participants throughout the study and about how to protect their sensitive data without undermining its richness. As an important starting point, we commissioned an independent review of the case data we were considering for inclusion in the showcase archive by a journalist with no previous contact with the study and a commitment to making academic studies more widely accessible. This process convinced her of the need to anonymise this data for archiving (and confirmed our own instincts). What followed was, once again, a staged and labour intensive process:

 

Enhancing transcripts

Anonymising transcripts

Tracking changes

Cross-checking and standardization

Researcher process notes

Negotiating a break with anonymity

 

 

Enhancing transcripts

View/Download: Transcription guidelines

Transcription can be an inaccurate process, complicated by a range of factors including the (in)ability to understand different accents and dialects. Re-listening to tapes to rectify serious errors or gaps has always been an important part of the qualitative research process. However, we found a further need to ‘clean up’ our transcripts. After each round of interviews, audio tapes were transcribed according to a set of guidelines we devised in 1998. These involved a very rudimentary mark-up system that would, for example, make real names readily identifiable. A number of transcribers were used over the ten years of the study with the result that there was some variation in the way in which the guidelines were followed.

 

Over time, we came to describe this process of restoring the transcripts to an accurate record of the audio tape and standardizing mark-up as ‘enhancing’. This was a result of presenting our work in seminars and conferences and getting feedback that the term we used previously - ‘cleaning’ - led to an assumption that we had somehow changed and impoverished our data.

 

(top of page)

 

 

 

Anonymising transcripts

 View/Download: Anonymisation guidelines

When publishing our own analyses and interpretations of the data, we used the normal devices of the pseudonym, the red herring or changing / omitting any identifying detail as anonymisation tools. This experience, together with a review of existing advice, served as a basis for developing an initial set of guidelines for anonymisation . However, preparing the data for others to use whilst maintaining the fine balancing act between keeping the richness and essence of the story and protecting the individual from unnecessary exposure proved much more complex in the QL context - since identifying detail is accumulated and considerable ‘flesh’ added to the ‘bones’ of each biography over time. An increasing awareness of the need to protect against this cumulative effect within any given case and across-cases within a particular research site meant we had to adapt our approach over time. Getting the whole picture of a case (all interviews) before anonymising and tuning into the significance of different people, places and events over time became essential. So much of what is of broad social interest and significance in a person’s narrative can lie in the fine detail and minutiae. In isolation, mention of a hobby or a place or a job or a family incident may not necessarily identify the person - but put all this information together and the jigsaw starts to build. Information picked up in interview 6 may influence the way in which certain facts are anonymised in interview 1. Their significance to the stories of other young people in the same research site also has to be considered and weighed. The picture keeps building as other interviews from the same research site line up on either side.

 

Specific issues in the QL context

Read more...

 

(top of page)

 

 

 

Tracking changes

View/Download: Tracking table

 

Carefully tracking all changes was important to us for two main reasons: the first was for our own reference purposes; the second related to use of the dataset in the distant future, when un-anonymised transcripts may be available. Basically, the user would need to publish data in the anonymised form – and would have a ready-made guide for doing this.

 

The tracking process involved logging the identity number for the relevant young person, the interview number, the page number of the change, the original and the change it had undergone as well as logging any possible future changes. XML-based mark-up tools were in an early stage of development at this time and consequently we worked in Microsoft Word. The decision to log page numbers – for easy reference – threw up technical issues (how to ensure everyone who accessed the transcripts would be presented with the corresponding page number). These were resolved by standardising and recording the page formatting (see ‘Anonymisation Guidelines’).

 

(top of page)

 

 

 

Cross-checking and standardization

View/Download: Cross checking notes

 

Close communication and comparison between anonymisers was essential to the anonymisation process. Once all cases were enhanced , anonymised, each case was thoroughly cross-checked by another team member (usually, the original interviewer). Finally, one researcher reviewed all enhanced transcripts and all enhanced and anonymised transcripts. This involved a ‘speed read’ to identify any anonmalies that had slipped through the net: e.g. real names, incorrect fonts and margins, hangovers from previous method of mark-up, and researcher’ notes or queries left in the body of the text. It was also a final opportunity to make final decisions on outstanding queries: some were taken to the team meeting, others discussed with the relevant interviewer.

 

(top of page)

 

 

 

Researcher Process Notes

View/Download: An example of researcher process notes

 

Capturing the anonymisation process as it developed over time was an important consideration for us and team member conducting the work kept process notes as they went along. 

 

(top of page)

 

 

 

Negotiating a break with anonymity

At the same time as we were developing our archiving strategy, we were also expanding on another model for re-use of the Inventing Adulthoods dataset: one in which we, the research team, maintained more control over the data sharing process by working collaboratively with others. In collaboration with a course team and a film maker at the Open University, we produced a course dvd in which five young people from the study were filmed: talking about their lives and their involvement in the study, and listening to and reflecting on their interview tapes. In these short films, the young people’s anonymity was no longer protected (e.g. by their study pseudonym) – as they used their real names and revealed details about their lives. This departure meant re-negotiating with them the basis for their informed consent – a process that continues as we use short clips from these films in an increasing range of contexts to illustrate our work on both the study and the archive.

 

(top of page)