I think the problem of testing for randomness of these suicides is harder to solve than it seems. Randomness only occurs in the absence of suicide letters which state that the leak caused them to do it. At least two scenarios:
1. They leave suicide notes where they explicitly attribute it to the hack. Then, the probability that it is due to the Ashley Madison hack becomes = 1. The question then becomes, did the 'outing' push them over the edge. Especially so, if they attribute taking their life to multiple reasons.
2. If it is not attributed to the 'outing', then one has to carefully see what the probabilities of suicide are in their particular cross-section of population and also consider other risk factors. The numbers someone else quoted here, 16/100,000 is an average. It is fallacious to assume that this number uniformly applicable to the population as a whole. You have to consider prior probabilities for each individual. These priors can possibly be computed based on the individual's medical history, socio-economic background, geographical location, etc. with importance assigned in that order. I would imagine medical conditions would play a big role. For example, if an 'outed' person with no history of depression, etc. from a hyper-religious family commits suicide, it is quite likely that it was because of the 'outing' in the absence of explicit attribution in a suicide letter, etc. If a person has on his/her history previous suicide attempts, depression, etc. the marginal probability that it was solely due to the Ashley Madison outing would probably be lower.
Comments
I think the problem of testing for randomness of these suicides is harder to solve than it seems. Randomness only occurs in the absence of suicide letters which state that the leak caused them to do it. At least two scenarios:
1. They leave suicide notes where they explicitly attribute it to the hack. Then, the probability that it is due to the Ashley Madison hack becomes = 1. The question then becomes, did the 'outing' push them over the edge. Especially so, if they attribute taking their life to multiple reasons.
2. If it is not attributed to the 'outing', then one has to carefully see what the probabilities of suicide are in their particular cross-section of population and also consider other risk factors. The numbers someone else quoted here, 16/100,000 is an average. It is fallacious to assume that this number uniformly applicable to the population as a whole. You have to consider prior probabilities for each individual. These priors can possibly be computed based on the individual's medical history, socio-economic background, geographical location, etc. with importance assigned in that order. I would imagine medical conditions would play a big role. For example, if an 'outed' person with no history of depression, etc. from a hyper-religious family commits suicide, it is quite likely that it was because of the 'outing' in the absence of explicit attribution in a suicide letter, etc. If a person has on his/her history previous suicide attempts, depression, etc. the marginal probability that it was solely due to the Ashley Madison outing would probably be lower.