Would it be feasible, presently or in the future, to use computers with blockchain technology to help avoid researcher bias?

For instance, in a medical experiment, when an experimenter uses blockchain to store data gathered by interviewers about a patient's condition after a new medication, he/she would be able to guarantee that it would be literally impossible to change the results to fit the expected outcome.

  • 3
    When people talking about others fudging results, it's not usually the veracity of the data itself that is in question; rather it's researchers who -- through malice or misunderstanding -- use questionable statistical methods. Straight-up falsifying data is rare, p-hacking, less so. – Azor Ahai Jun 12 '19 at 18:20
  • 2
    It might be a solution for changing data, but how would it avoid bias? Bias can occur immediately when the data is collected and entered. Even automatically collected data can be altered if you control the software that collects it. – Buffy Jun 12 '19 at 18:22
  • 3
    Slightly relevant: xkcd.com/2030 – Buffy Jun 12 '19 at 18:26
  • I was thinking about a case when researcher intentionally use p-hacking or data falsification from accurate results collected by interviewers. So blockchain might be able to prevent data from being manipulated by researchers in a foolproof way. – Jay Lee Jun 12 '19 at 18:26
  • 5
    Why blockchain? That's complete overkill. Upload your datasets onto your university's public archiving system. Everything will be signed and versioned and that's it. – Karl Jun 12 '19 at 18:47

While blockchain might be a good solution for certain (future?) applications in research, experiments, and data collection – it is hard to shine a lot of light on it right now, as the technology is pretty new. And research techniques, scientific methodology, experiment design, and, most importantly, ethical standards have been (and will be) developed for centuries.

Most importantly, a lot of interaction between scientists is (and should be) based on trust and an assumption of good intentions.

This particular question focuses on researcher bias; however, the example is more about data manipulation or "forced outcome" research.

Regarding data-manipulation:

  • As correctly mentioned in the comments (@AzorAhai), it is rare to forge the raw data itself, but use certain statistical techniques to get to the desired outcome.
  • Using blockchain would have to be introduced on a lot of levels: not necessarily just the surveys. Blockchain for measurement equipment (which constitutes a lot of experimental data in natural sciences) is not yet there. And if one allows a researcher (how crazy is that!) to upload measurement data himself to blockchain-based storage – nothing prevents the researcher from forging data before uploading it.
  • Not being able to change the data might not always be a good thing. Sometimes, you have to change it (simple use – delete certain entries) due to technical (it was found that the equipment was broken) or ethical reasons. Here the "permanency" advantage of blockchain might actually be damaging. For example, even git (version-control system not allowing commit change without the change of the signature) has some widely discussed issues: ACCU talk of Alex Chan, the particular segment on GIT and personal info issues.

So, regarding the experiment data, I currently do not see the advantages of using blockchain anywhere in the research loop.

I can see some potential use of blockchain in:

  • review process (though, I would say it's an overkill) to anonymize authors/reviewers/editors. However, there are other simpler tools to solve it – and even they are rarely used in practice.
  • authorship proof – still there are other tools.

To conclude:

I am very interested to see practical uses of blockchain in other areas (let's see where the smart contracts go in 5-10 years). Academia and research process does not scream for blockchain and its usability is totally unclear and hard to implement.

| improve this answer | |

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