Understanding Research Ethics: The Challenges of Going Undercover

Last time, we talked about how a researcher can blend in with the crowd.

We discussed the Visual Assimilation Protocol, which helps someone look normal in a strange place. This helps a researcher remain unnoticed while working in a sensitive area.

That idea needs more discussion.

A method that asks a researcher to play a role brings three must-follow rules: the role should be controlled, the researcher should be ready to leave, and the trick must be ethically right.

The last one is the toughest.

I will explain them one by one.

The Persona Is a Contract, Not a Costume

Creating a cover is the easy part.

Carrying it out is the real challenge.

Where are you from? Who do you know here? Why did you leave your last job? How long have you been doing this? Who taught you?

Every answer is a claim. And each claim can break the cover.

There are three main rules to follow.

Keep the persona simple. The best cover story is a short one. Every made-up detail is another fact to remember, and more facts mean more chances to get caught. A persona should be as close to the researcher’s real life as possible, with little fiction.

Do the work right. A researcher working in a shop or market must do the job properly. The persona is not just an excuse to be there. It is a duty to the people around. If a researcher pretends to be a vulcaniser's assistant but does it poorly, they are not just hurting their research. They are also being a bad neighbour.

Set rules before stepping into the field. There are actions the persona must avoid, no matter the situation.

It must not take part in illegal activities.

It must not give information to armed groups or violent people.

It must not use friendships or close relationships just to gather information.

And it must never pretend to be someone else by using their name, documents, or reputation.

These limits matter more than how clever the technique is.

When the persona starts making decisions for the researcher, the method becomes dangerous.

The Exit Is Planned Before the Entry

In Part 2, I mentioned a researcher who blended in well.

But I didn’t stress enough that being able to leave is part of the plan.

At the Sundiata Post Knowledge Lab (SPKL), the main rule is simple:

No researcher enters a zone without a written exit plan.

This means four things.

Someone outside knows the plan. A main contact has the schedule, the check-in times, and the process if check-in fails. Immersion is not a solo job.

Triggers are set in advance. The point when an assignment ends must be clear before the researcher gets too involved. This includes any real threat to life, which the rules below say takes priority over the persona. The conditions that end the assignment are agreed calmly, as they might not be easy to assess in the field.

The exit should be normal. A sudden exit can draw attention to the researcher and the people who worked with the persona. The best exit has a believable reason and leaves the community with little risk. That ordinary reason is another small lie, and it must stay consistent for as long as the people who trusted the persona are around.

The data should leave separately from the person. Field notes should be minimal. When possible, they should be written after leaving, not while still on site. Information should be stored so that losing the researcher’s device does not put others at risk, and whatever is recorded should be safe if lost. After leaving, the researcher writes a record of what was done, how, and at what cost, without revealing identities, for review.

Under these four rules is one big question:

Who would be blamed after I am gone?

The researcher leaves. The market stays. The neighbours stay.

The people who trusted the persona remain. That is why the exit is not just about logistics. It is about ethics.

The Ethics of Research in the Dark

Now I will discuss the third obligation.

By the ethics of going dark, I mean the responsibilities of a researcher using a fake identity, without the people being watched knowing, in places where being open would be unsafe or would ruin the research.

It asks two questions. When, if ever, is hiding justified? And what do we owe to the people involved?

The answer has four parts: tests to decide if going dark is okay, rules for harm seen or faced, honest sharing of the method later, and duties to both the people being watched and the researcher.

This is not new. Research ethics has debated covert methods for years, and this column builds on that discussion instead of replacing it. Journalism has talked about this for generations. Neither has decided that trickery is never okay. Both agree that it is not automatic.

Deep immersion is highly valued, but it is only as good as the ethics behind it.

Deep immersion can give the best data. That is exactly why it faces the toughest tests. Four tests apply.

Necessity. Immersion is the last option in the Tri-Tiered Approach. Remote tracking comes first, then rapid intercept. If either can answer the question, immersion is not necessary.

Proportionality. The knowledge gained must justify the deception used to get it. The level of risk rises with the vulnerability of the people being watched. Studying a market seller is harder to justify than studying an official. Foreseeable harm means any harm someone familiar with the community could expect: retaliation, stigma, lost jobs, or legal trouble. No one being watched should face harm just so the researcher can get better data.

Protection. Names, specific locations, and identifying details should stay out of the records unless absolutely needed. Data that cannot be kept safe should not be collected just because it seems interesting.

Accountability. No group should approve its own secret research. Before any immersion, an independent panel of outside advisers reviews the case in writing. The panel’s rules are made public. They can veto. Their decision is included in the method account. And no researcher enters a place without a legal review for that area.

Three tough cases test these rules harder than any others.

What if the persona sees a crime?

The rule that the persona must not take part in illegal activities covers actions. It does not cover what the researcher observes.

Reporting may put the people the researcher is protecting in danger. Staying silent may make the researcher complicit. The rule must be set before entry, not at the moment.

Life is more important than the cover. If a life is in imminent danger, meaning likely to be lost before help or withdrawal could prevent it, the researcher must warn or leave, even at the cost of the persona. Everything else is recorded, and the panel will decide what happens next.

A researcher who has not settled this beforehand will be forced to decide under pressure. That is the worst time to make such decisions.

What if the persona is exposed?

Exposure, detention, and compromise are not unlikely events. They are possibilities the protocol prepares for.

The researcher must withdraw. The main contact informs the panel. The first step is to protect those who worked with the persona. The failure is reported in the method account like any other outcome.

What if revealing the truth puts people in danger?

I said the deception should end, at the latest, when the findings go public. But publication can expose the very neighbours the method aimed to protect.

So revealing the method has two parts. The method is fully disclosed: what was done, how, why, and at what cost. The identities are not shared. Details that could link a finding to a stall, street, or person are changed or withheld, and the account states this openly.

The community’s consent cannot be sought beforehand, because asking would defeat the research. That is why the other safeguards must be strict. Where consent is impossible, restraint must be the priority.

The Duty to the People Observed

The researcher’s protection is only part of the responsibility. The people being observed also face risks: retaliation, stigma, broken trust, and legal issues. They get none of the pay, training, or recognition.

So no one being watched should face foreseeable consequences for the findings. If they find out about the research, they can respond through someone outside the unit. Findings are returned to the community in a way they can use. And no benefits should be taken from the research without sharing it.

The Duty to the Researcher

An institution should never ask someone to take a risk it would not take itself. Field researchers need to be trained, briefed, paid fairly, and allowed to refuse or withdraw without penalty.

A method that works only because someone else bears the risk is not a true method.

It is just passing on the danger.

That applies to the researcher. It is also true for the community.

SPKL will publish its field ethics protocol, covering legal reviews, data handling, and the panel’s rules, so that every promise here can be checked.

The Lie Must End. The Record Must Not.

Part 2 ended with the researcher living "a quiet lie" to uncover the truth.

I still stand by that phrase. But the tests, the panel, and the rules for tough cases all lead to one condition: a lie in the field is only acceptable if the account later is honest about the method.

That means being truthful about what was done, how it was done, why it was done, and what it cost.

The people may be protected. The method may not.

That is the difference between a researcher who deceives and a fraud who just writes notes. The first explains the deception. The second hides it.

The wardrobe was never the main point. It was just the first part of a much bigger task.

The researcher must master the persona. The researcher must plan the exit. The researcher must pass the tests. And finally, the researcher must be ready to explain the whole process.

Going dark does not mean avoiding accountability. It means carrying accountability into the dark.

Part 4 will look at how a stall in a market, a checkpoint on a road, and a chat over a recharge card or airtime become a number on the Decoupling Sovereignty Index.

Field evidence is only useful if it can be turned into a system that can be scored, tested, and challenged.

The researcher may disappear into the field. But the method cannot.