Background

Questioninfo icon

How did Mahsa Amini die?

11 Dec, 2022

Answerinfo icon

Amini died as a result of a heart attack or stroke (not induced by violence).

(71% probability)

Backgroundinfo icon

On 13 September 2022, a 22-year-old Iranian woman named Mahsa Amini, was detained by Iran’s Guidance Patrol (also known as “the morality police”) for violating Iran’s dress code. Two hours later she was hospitalized, and after three days in a coma, she was declared dead.

Her death sparked a wave of protests in Iran by those who believe she was murdered by the police, while the official investigation claims she died of natural causes. As no definitive evidence emerged thus far, a probabilistic analysis is needed.

Hypotheses Consideredinfo icon

Calculated Resultsinfo icon

Calculated Resultsinfo icon

1

71%
Natural Causes:

Amini died as a result of a heart attack or stroke (not induced by violence).

71%

2

15%
Involuntary Manslaughter:

Amini was struck by the Guidance Patrol, leading to an unlikely nor intended outcome.

15%

3

14%
Murder:

Amini was intentionally beaten to death by the Guidance Patrol, for example for religious reasons.

14%

Starting Pointinfo icon

Initial Probabilities

Name

Updated Likelihoods
Murder
50%
Involuntary Manslaughter
50%
Natural Causes
0.9%

To calculate the initial probabilities of the Murder and Involuntary Manslaughter hypotheses, we researched all deaths that occurred in the last 20 years while in the Guidance Patrol’s custody (in circumstances similar to Amini’s) and divided this number, by the number of people they detain.

For the Natural Causes hypothesis, we calculated the combined likelihood of a person in Amini’s age group dying due to her claimed causes of death (a stroke or heart attack).

While dying while in the Guidance Patrol’s custody is rare (we found only two cases that are somewhat similar to Amini's), dying at Amini's age from a stroke or heart attack is even rarer. This means the Natural Causes hypothesis is highly unlikely compared to the two competing hypotheses, before considering the specific evidence in this case.


Name

Updated Likelihoods
Murder
50%
Involuntary Manslaughter
50%
Natural Causes
0.9%

Evidenceinfo icon

Effectinfo icon

Video of the incident

Name

Effectinfo icon

Updated Likelihoods

Murder
÷1000
2.4%
Involuntary Manslaughter
÷350
2.4%
Natural Causes
×1
95%

The CCTV video documenting Amini from her arrival at the detention center up until her collapse depicts no violence and no indications of prior violence. Amini and everyone who exits the police van with her seems calm and acts normally. 

As such, this evidence is extremely unlikely under the Murder hypothesis, somewhat unlikely under the Involuntary Manslaughter hypothesis, and to be expected under the Natural Causes hypothesis. 

Name

Effectinfo icon

Updated Likelihoods

Murder
÷1000
2.4%
Involuntary Manslaughter
÷350
2.4%
Natural Causes
×1
95%

Medical diagnosis

Name

Effectinfo icon

Updated Likelihoods

Murder
÷6
0.5%
Involuntary Manslaughter
÷2.2
1.1%
Natural Causes
×1
98%

There have been three main sources claiming to provide medical evidence or a diagnosis as to the cause of Amini's death: leaked medical tests that appeared in the media, pictures of Amini in the hospital, and an official diagnosis by Iran’s coroner.

We disregard these claimed tests as they are uncorroborated and originate from an unreliable source. Amini’s hospital pictures and the coroner report show no indication of violence and as such, reduce the likelihood of the Murder and Involuntary Manslaughter hypotheses.

Name

Effectinfo icon

Updated Likelihoods

Murder
÷6
0.5%
Involuntary Manslaughter
÷2.2
1.1%
Natural Causes
×1
98%

Testimonies

Name

Effectinfo icon

Updated Likelihoods

Murder
÷43
0.5%
Involuntary Manslaughter
÷43
1.1%
Natural Causes
÷16
98%

Testimonies appearing in the media have mostly described Amini being beaten by the police, which supports the Murder and Involuntary Manslaughter Hypotheses. However, this effect is reduced as all of these testimonies are uncorroborated, hearsay testimonies, mediated either by media organizations funded by countries aiming for regime change in Iran or by Amini’s family, who are more likely to provide such information whether true or false. At least part of Amini’s family also has a significant bias against Iran’s government and is active in a political and armed conflict against it, with the hope that "these protests will continue and it will end with the fall of the Islamic Republic". On top of this, some testimonies contain internal contradictions while others are contradicted by the CCTV footage. 

 

A few of the testimonies do carry more significant weight: Amini’s father claimed she did not have brain surgery (contrary to official claims) and that he saw blood on her body. These claims further support the Murder and Involuntary Manslaughter hypotheses while one testimony significantly supports the Natural Causes hypothesis: Amini’s brother received the diagnosis that his sister suffered a stroke and a heart attack within 17 minutes of her collapse. This is unlikely to be sufficient time to establish a conspiracy to hide her true cause of death and thus was likely a truthful diagnosis.  

 

Name

Effectinfo icon

Updated Likelihoods

Murder
÷43
0.5%
Involuntary Manslaughter
÷43
1.1%
Natural Causes
÷16
98%

Ignored Evidence

Discussioninfo icon

userIcon
user icon
Brandon
Apr 21, 2023 at 9:47 PM
I think I found a mistake! In "The likelihood of death by natural causes", you give an annual rate of stroke, but Amini was not in custody for that long — she was in for much shorter. This makes it less likely she died of a stroke while in custody than the above numbers would indicate. Could you account for this?
user icon
David
Dec 30, 2023 at 11:55 PM
While it's only a small shift in probabilities from 87% to 71%, it's a larger shift in odds from "6.7 times more likely" to "2.4 times more likely" – the odds are now almost 3 times lower. Anyway, I'm impressed by the rigour of the analysis, and especially by the transparency and honesty! 6.7 = 0.87/(1-0.87); 2.4 = 0.71/(1-0.71)
logo icon
Rootclaim
Jun 1, 2023 at 7:35 AM
After further review, we found there was indeed an error in our analysis. We fixed it, shifting the leading hypothesis from 87% to 71% probability. While the error was significant, the difference in the final result was relatively small, since our model already accounts for such mistakes. Congratulations on being the first person to find a significant error in a Rootclaim analysis.