The disciplinary notice kept Wema out of the pantry for two days.
On the first day, she spread every relevant record across a newsroom table: consent forms, the anonymous matrix, radio-call logs without caller identities, the ethics ticket, and the reporting methodology she had written before the assembly.
Halima pointed at the folders. “We give them everything.”
“Not everything. Source identities stay protected.”
“What if they say they need them?”
“Then they show us the policy that gives them that authority.”
Professor Lillian Muthoni, the student press ethics adviser, arrived at noon and began with questions that annoyed Wema because they were exactly the questions an investigator should ask.
“Did you collect any datum without consent?”
“No.”
“Did you access another student’s institutional account?”
“No.”
“Do you possess screenshots of private accounts?”
“No.”
“Did you ever see decision letters?”
“Yes. Participants showed some voluntarily. I recorded only the fields needed for the process comparison unless they separately authorized a redacted copy.”
“Good. Do not defend yourself by pretending you were perfect.”
They built an evidence index. Every item had a source, consent status, and purpose.
Item one: Wema’s own data export. Item two: public governance documents. Item three: consented participant testimony. Item four: the anonymized process matrix. Item five: the radio consent log. Item six: the council correction. Item seven: confirmation of the protected ethics inquiry.
There was no raw scholarship database.
Wema also wrote out the limitations of her reporting. The participants were not a random sample. Some contexts were self-reported. The team did not have source code. A few rows previously described as “false positives” were more accurately “potential false-positive candidates.”
She corrected that wording on the student press site before anyone demanded it.
Halima stared at the update. “No one was going to notice that phrase.”
“I noticed.”
“That is exhausting.”
“Yes.”
At the disciplinary interview, the dean sat with a legal officer and a Student Affairs representative.
“Miss Kilonzo,” the dean began, “we have concerns that you aggregated restricted student records without authorization.”
“I aggregated consented testimony and process outcomes. I did not access the university’s restricted record system.”
“Your notebooks?”
“I can provide methodology, consent templates, and redacted evidence chains. I will not provide protected source identities without a valid requirement under the student media policy.”
The legal officer asked, “Aggregation itself can create re-identification risk. Did you consider that?”
“Yes. That is why I removed combinations of exact date, course, residence, and personal financial detail where they were not necessary.”
She handed over the matrix.
There were no names, phone numbers, addresses, family incomes, or photographs.
The dean studied it.
“Who gave you these records?”
“The participants supplied testimony and showed their own decisions voluntarily. The ethics inquiry record is not mine.”
Professor Muthoni placed the inquiry confirmation on the table. “An institutional disclosure exists independently in the protected channel.”
The accusation changed shape.
The legal officer leaned back. “I do not see evidence here of unauthorized database access.”
Wema did not smile.
“And the pantry suspension?”
“Pending final review,” the dean said.
“Why is a food-pantry role linked to this allegation?”
“Risk management.”
Wema wrote that phrase down.
She left without a dramatic exoneration, but the central accusation—data theft—had lost its evidentiary basis.
Back in the newsroom, Professor Muthoni told her, “The pantry suspension may persist because bureaucracy moves slowly. Do not automatically label every delay retaliation.”
Wema nodded. “I will distinguish the action from the motive.”
Then she published her full methodology.
It described consent, minimization, field definitions, uncertainty, counterexamples, and the exact point where observation became inference. It also included a corrections form.
Within an hour, someone found a counting error in the travel category. Wema fixed it and left a revision note.
“This is the price of open method,” Halima said.
“It is the price I asked the university to pay too.”
Students from the statistics department contacted her that evening.
*We can blind-code context and compare it with outcomes using new volunteers.*
Wema stared at the message.
An independent test could strengthen her work.
It could also prove her pattern was weaker than she believed.
She replied:
*Yes. Keep consent and data minimization. Publish limitations even if the result disagrees with me.*
By making her method public, Wema had gained legitimacy and lost monopoly.
Publishing the methodology also forced Wema to accept corrections from people who disliked her conclusions. One reader noticed that an earlier article had used the phrase “false positive” for a case whose context was only self-reported. Wema changed the phrase to “potential false-positive candidate” and left the revision visible. The correction was small, but it demonstrated the standard she wanted the university to adopt: records should show how claims change when better evidence arrives. The statistics students who later offered the shadow audit told her that the open correction made them more willing to collaborate, because it suggested she would not bury a result that weakened her story.
The investigation no longer belonged only to her.