Can you publish studies based on confidential customer data that comes from private companies?

Question

I am working on a project in biostatistics involving some private companies. Specifically, these companies share data about their customers with us and we try to apply our algorithm to this data. The good part is that both the algorithm and the application are very new, and seem to work quite well. Unfortunately we can not publish this data in paper or conferences, as it is privately owned.

• What is the typical way to proceed in this case?
• Does the fact that it is not possible to publish the data implies in general that also results based on this data are also unpublishable?

Answer 1

Publishing data based on confidential private company data Where the company consents, it is possible to publish studies based on confidential customer data. Common requirements are

• Don't identify the company in any publication and possibly be broad when discussing the context of the company (e.g., location)
• Don't make the raw data available to other researchers or on online repositories
• Don't present individual-level analyses or small group analyses which might allow any customers or proprietary data to be identified

When you initiate a relationship with a private company to analyse their data, it is important to discuss and agree on a number of matters:

• Are you able to publish research based on analyses of their data? (note that this can be done without sharing their raw data)
• If you publish, to what extent do they require their details to be concealed? (e.g., no disclosure of place, companies, etc.)
• How will authorship of publications work? Does anyone at the company wish to be a co-author? If so, what level of contribution do they need to make and how will this work?
• What if any rights of veto or influence do they have on what can be published? Will you have academic freedom to publish results even if they do not support the company's commercial agenda?
• Are you able to share their data with other researchers or on suitable online repositories?
• Will they give you money or other compensation for analyses you perform or reports you write for them? If so, what are the details?

Types of Relationships:

In general, there are different kinds of partnerships between academia and companies that relate to the analysis of data. Relationships should be formed based on a shared understanding of goals and mutual benefit.

• Pure consulting: In this context, while you are in academia, you can often still provide standard consulting services. In this setting, you would charge commercial rates to the company for any insights you gain.
• Pure research: Data is shared with you by the company with the explicit aim of assisting your program of research. The company may be curious about the results, but is principally motivated by a general desire to assist in your research. The company does not have any explicit expectations about customised reports and so on beyond the research program that you pursue.
• Hybrid consulting/research: In this model the company is seeking to gain insights into their business. Providing customised reports to them and other services may be part of the exchange for allowing you access to the data for research purposes. Alternatively, allowing you to publish research based on the data may be one of your conditions that may offset the fee you charge the company for other services you provide.

Within all these models, there are many variations. The important point is that early on in the relationship, you ensure that both you and the company agree on what model is being adopted. These matters can also be renegotiated.

Answer 2

It will very much depend on the data in question, and what agreements you've reached with the private companies. For example, a lab I used to work out paid rather handsomely for some data that underlies some of our models, but it's basically ours. We publish as we see fit.

A couple points:

• If you want to publish the data itself, that should have been negotiated in advance, though it's reasonable for a company to say no.
• If you wish to proceed, you could try to renegotiate, but again, this should have been done ahead of time.
• No, not being able to publish the data does not imply you cannot publish the results of your research. You should consult the original agreement made with this company to see if, and how, publishing results is allowable.

If you genuinely can't get anywhere, it may be possible to simulate data with similar properties and publish on that.

Answer 3

I won't do it, the company where I did my research allowed me to publish the results of investigation, but they warned me about future workers that may not be happy with the publications. Anyways, if you dontt allow any reader to get an idea of which is the company, there wont be any problem.

But talk with the workers of the company first.