6

Background:

I am writing a paper analysing various organisations that I need to de-identify.

For my analysis I need to cite three types of sources:

  1. Publicly available published works that identify the organisations (e.g. reports by the organisation, peer reviewed works that assess the performance of the organisation, etc)

  2. Interviews I have conducted with the organisations where I need to de-identify the organisations.

  3. Personal correspondence (emails) with the organisations where I need to de-identify the organisations.

For the sake of explanation, lets pretend that "Organisation A" is Google (in actuality this report has nothing to do with Google). I have undertaken an interview with Google (that needs to be de-identified per ethics approval), had personal correspondence by email with Google (that needs to be de-identified) and also I want to cite public reports Google has published.

The problems:

When combining these three types of sources I end up writing sentences like this:

"Organisation A performed well against validity metrics (interview with Organisation A), sufficiently against accuracy metrics (personal correspondence with Organisation A) but poorly against correctness metrics (Google et al., 2020).

Problems with that include:

  1. When I am analysing publicly available sources alongside interviews I essentially re-identify the organisation. It is obvious in the above sentence that Organisation A is Google.

  2. These sentences are really ugly with all the interjections of "(personal correspondence with Organisation A, 2021)" and "(interview with Organisation A, 2021)".

The questions:

  1. How should I deal with citing these three types of sources without re-identifying organisations?

  2. Is there a citation style that will make this less ugly? Maybe something where personal correspondence and interviews can be references with a superscript or similar? At the moment I am using Vancouver and I don't think it is the best choice.

Thank you kindly for your assistance.

Edit: clarification - even though I know that it is not necessary to reference your own interviews in the text of your study (i.e. interviews that have been undertaken pursuant to a description in the methods section), I would still like to find a notation style that allows me to intermingle interviews, personal correspondence and public documents in a way that is visually clean but also shows where information has come from (i.e. not ugly in the way I have described above). This issue is almost separate and apart from the issue of deidentifying sources.

9
  • 2
    What did the IRB that approved your interviews suggest?
    – Bryan Krause
    Jul 14 at 23:26
  • 1
    Ok. There's typically no need to cite your own research in your writing using a "citation style" - that's for citing other work. You might use it if you were referring to interviews that are published in some way, of course. Your methodology should make clear that you've used interviews to draw your conclusions. But you need to work through your IRB on this - whatever suggestions you might get here are still things you need to run by IRB if they've decided things must be de-identified. Probably you can't make comparisons like you want to if you don't have approval to release the IDs, though...
    – Bryan Krause
    Jul 14 at 23:33
  • 1
    (not even without connecting specific Org A to Org A... even if you just use Org 1 2 3 for the public data and Org A B C for the interviews, even being able to know that the orgs are the same identifies them)
    – Bryan Krause
    Jul 14 at 23:35
  • 1
    Yeah, I will certainly be running everything past the relevant IRB, I just want to see how other people would approach this problem to get some ideas. Jul 14 at 23:39
  • 1
    Okay, great. We frequently get people asking here that seem to have absolutely no idea what they are doing with human subjects stuff so I always get a bit nervous about it because of how much trouble people can get into when they get ahead of themselves. I think your question (2) is very easy, but your question (1) may not have any good answer. Good luck!
    – Bryan Krause
    Jul 14 at 23:41
1

You don't need to cite your "data" every time you mention it

Assuming it is okay to identify the institutions in your study, I would recommend that you cite the published reports and other public works from these organisations (using standard citations), but for the remaining material you can consider this as part of your "data" and you can just mention how it was collected in your methodology and not cite it each time. The results of the interviews you conducted, and any correspondence dealing with follow-up questions are essentially just the recorded "data" for your project. You should save this material in an appropriate form with your institution, and it should be available to researchers on an appropriate basis having regard to privacy, etc. (which might preclude access).

For example, in your methodology section you would describe the interviews you conducted in sufficient detail for the reader to understand the methodology; it would then go without saying that the results in the interviews are recorded in your data for the project and you would have no need to cite any source when you set out the results of the interviews.

If you want to completely de-identify the institutions then you would not give citations to their policy documents, but you would note that internal policy documents were the source of the metrics (and in your internal data you would record the sources of each of the metrics for your own use). Alternatively, if it is okay to identify the organisations in the study, but you want to de-identify their results (so you give results that are not matched with organisations), you would cite their materials in your methodology section to establish the relevant metrics, etc., but you would then de-identify in your results; you would have no need to re-cite those sources there. If the internal policy documents are quite voluminous (and I can imagine they would be) you could put your citations in the methodology section in a table. For example, you might do something like this:

Methodology

We did a series of interviews with employees at ten institutions: Google, Apple, IBM, ... . Performance metrics were taken from internal policy documents at those institutions; see Table X for sources for each institution. Questions for our interviews are shown in Appendix B; these were supplemented by correspondence to clarify policies, metrics, etc., where needed.

Then later...

Results

For purposes of de-identification we have randomly assigned our organisations as Organisations A-J, and they are described with these labels in the results and discussion. ...

Organisation A performed well against validity metrics, sufficiently against accuracy metrics, but poorly against correctness metrics.

0

I agree with the responses above. Once you identified and cited the sources of the data, statistical/qualitative results don't need citations because you found them. In terms of the citation style, I think you are asking the wrong question. Whatever journal you plan to submit the article, you should follow their instruction about the style.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.