Political Candidate Deaths: Seven German AfD politicians die within weeks of each other ahead of elections

What are the chances? A lot higher than you think.

4 Sep 2025, Breitbart: Seven AfD Candidates Die Ahead of Local German Elections

Several candidates for the populist Alternative for Germany (AfD) have died in the lead-up to this month’s local elections in the country’s most populous state.

According to the German paper of record Die Welt, at least seven AfD candidates have died ahead of the September 14th elections in the state of North Rhine-Westphalia. Authorities have stressed that there is so far no evidence of foul play in any of the deaths.

Germany’s Deutsche Presse-Agentur wire service noted that a total of 16 candidates have died before the upcoming vote. Yet, no other party besides the Eurosceptic right-wing party has suffered more than one death.

However, police have already confirmed natural causes in four of the AfD candidate deaths and said there are so far no indications of foul play in the others.

The latest death was confirmed this week to be Hans-Joachim Kind, a candidate for the populist party in the Kremenholl district. It has been reported that the 80-year-old died of natural causes after a lengthy illness.

An 80-year-old man?! Clearly, that’s suspicious!

But more seriously, let’s look at the claims re: probabilities and see if they even approached the calculations correctly (and there are some missing pieces of info but that still won’t stop me).

Claims of Suspiciously Low Probability Event(s)

Alice Weidel is one of the leaders of AfD, and she retweeted this:

4 people, you say? (This is clearly before the numbers got up to 7.)

Let’s see the ages listed:

These were all in German originally.

While Stefan Homburg says “No typical age of death”, sorry, if they had died age 25, sure, I would agree. Dying suddenly in middle-age/early seniorhood is actually pretty common. Especially if it were from heart issues.

Or if they had cancer.

Gathering Some Important Numbers: Number of Candidates, Average Ages, Election Window

2 Sep 2025, BBC: Germany's far-right AfD suffers series of candidate deaths ahead of local vote

As many as six candidates for Germany's far-right AfD have died in recent weeks ahead of local elections in the big western state of North Rhine-Westphalia.

Police have made clear there is no evidence of foul play in the deaths, but it means that new ballots will have to be printed and some postal voters will have to recast their ballots.

North Rhine-Westphalia has a population of 18 million and a reported 20,000 candidates will run for office in its 14 September local elections.

The number of deaths has nevertheless raised questions on social media. The state's interior ministry has pointed out that candidates from other parties, including the Greens and Social Democrats, have also died.

….

Initial reports centred on news that four of its candidates had died, and then the deaths of two reserve candidates also emerged, prompting a flurry of conspiracy theories on social media.

….

Police told Germany's DPA news agency that the four initial deaths were either from natural causes or the cause was not being divulged for reasons of family privacy. The two further deaths have been similarly described.

First off, you can’t include the reserve candidates who died unless you include all the reserve candidates from all parties. So sorry not sorry, I’m throwing out those two deaths. I will explain that in a bit.

Here are the relevant pieces of information from the above we need to glean:

• 20,000 total candidates, there are about 6 major parties (but more do exist) and over 3500 seats in the election

• More than just the 5 AfD candidates died (again, excluding the two reserve candidates) - a few from other parties also died. I don’t have the full count, but we’ve got at least 7, it sounds like

• The full window, for simplicity, I’ll make 2 weeks

• Average age/sex combo of candidate I’ll make to be whatever makes the annual mortality rate to be 1% — I’ll show what that’s likely to be - and I’ll come back to this

There are reasons I need all of these items, because I worked on a problem like this for my 40th birthday.

Unlikely Events Occur All the Time

FOUR DEATHS OF (older) MIDDLE-AGED FOLKS IN TWO WEEKS!!!!

Do you throw a hissy fit at every bridge hand you see?

Because every specific bridge hand, if you specified it ahead of time is extremely unlikely, with it being one out of 635,013,559,600 possibilities.

That’s not fair! you cry — you have to get some hand, after all, when the cards are dealt. If you had called what the cards would be ahead of time, that would have been astronomically unlikely, but

There’s nothing that says that 4 people had to die out of the candidates in AfD in the specific period….but you didn’t specify that ahead of time, did you? Throwing in the deaths of two reserve candidates just make it more unlikely.

But people didn’t go digging for deaths of other reserve candidates, or other parties’ candidates.

This is just like a “psychic” doing a cold read, impressing the marks.

People have a very bad “feel” for probability, risk, and especially coincidences.

I remember getting into the whole ESP, “In Search Of…”, and the I WANT TO BELIEVE kind of gullibility of weird happenings….as a child. But it’s a natural reaction for humans. We like to find patterns, and we prefer deterministic cause-effect chains. We don’t like randomness.

But I’m about to show you how to estimate the probability here. It’s not what people have been doing.

Time to Calculate Some Probabilities!

I mentioned that for my 40th birthday I did a similar problem.

January 2014, CompAct, Society of Actuaries: Variations On Approximation – An Exploration in Calculation

In honor of my upcoming 40^th birthday, I am choosing as the situation to model the case of 2,000 40–year-old women and counting how many die in the next year.

Basically, any given individual had a very low probability of dying.

But if you have a large enough number of people, yeah, the expected number of deaths is going to be:

Getting my qx from the Social Security 2010 calendar year mortality table, I see the probability for an individual woman is .146 percent. Obviously, most of the probability mass will settle at the low end of the distribution, given that the expected number of dead women is 2.92 and the standard deviation is 1.71.

The point of the article was taking the “exact” distribution, which is binomial (assuming the deaths are independent, and the probability for each person is the same), and estimating the distribution in multiple ways.

To make this simple, for low probability events, one of the easier ways to estimate the distribution is the Poisson distribution.

This is the sort of thing actuaries do in setting reserves for insurance, by the way. We want to make an estimate of expected losses, and then we often look for a margin for safety. We will often look at a variety of data sources, but ultimately have to pick one estimate.

So let’s estimate the parameter!

Using the assumption set I had at the top:

First, let’s look at 1% per year mortality assumption:

Using the Human Mortality Database, and the German mortality 1x1 for calendar year 2019 (the most recent non-pandemic year for which they have data), the age/sex combos at which q_x (the 1-year mortality rates) are 1%:

• Male: 61 years old

• Female: 69 years old

Yes, there is a huge mortality sex gap in Germany, too. Probably smoking, related (and the East German aspects in there for the older folks.)

We can try some ranges on the probabilities here.

To calculate the Poisson parameter, there is an expected number of candidate deaths:

20,000 * 1% * 2/52 (52 weeks in a year)

= 7.692308

Here’s what the distribution for Poisson with expected value 7.69 looks like:

This is total deaths — so, we’d be expecting somewhere between 5-10 deaths in the bulk of the probability.

But here’s the deal - out of 20,000 candidates, I bet there were deaths that weren’t reported.

Did you hear about THIS CANDIDATE DEATH?

29 Aug 2025, North Salem Daily Voice: Acting Putnam Sheriff Nominated To Appear On Election Ballot After Predecessor's Death

Putnam County Executive Kevin Byrne announced the endorsement on Thursday, Aug. 28, praising Hess as the right leader to continue McConville’s legacy of public safety.

….

Hess, a veteran law enforcement leader with more than 25 years of experience, was appointed Undersheriff just days before McConville died peacefully at his Cold Spring home on Friday, Aug. 22, at the age of 68.

No, you didn’t. This is local to me, so I did.

Nobody goes around collecting statistics on the candidates who die in local elections. A lot of them die from normal stuff — heart attacks, car crashes, cancer, etc.

Nothing nefarious. It’s just the normal expected stuff.

Stress-Testing the Variables

Here’s the spreadsheet for you to download:

German Candidate Deaths Poisson Estimate 2025 09 05

35KB ∙ XLSX file

Download

Download

It’s very simple, but if you want to play around with it, you can expand the window (4 weeks) and drop the death probability (0.75%), and it will look up the equivalent ages for that death probability and graph the probability distribution.

If you adjust the parameters too much, you’ll need to adjust the graph - I have it graphing only up to 20 deaths. But I have the distribution calculate up to 99.

The highlighted area are the parameters to play with. I can go for younger average candidates, as here… but longer window of time.

And yes, this is the sort of thing I do for fun.

But I also more seriously did this in life reinsurance. I had to investigate “suspicious” deaths.

Below are some older posts related to the above - there really have been suspicious political deaths in the past. Such as Russian political opponents of Putin falling out of windows.

This isn’t it.

Related Posts

• March 2017: Mortality Monday: Suspicious Russian Deaths

• 19 June 2019: Mortality with Meep: How Many Deaths Before it's No Longer a Coincidence?

• 15 July 2019: Deaths in the Dominican Republic: How Many are We Up To Now?

• 20 June 2019: Mortality with Meep: Dominican Republic and Raw Death Rates

• 22 June 2019: Mortality with Meep: Even More Dominican Republic Deaths

• May 2023: Podcast: Mortality Trends and Cognitive Biases

• Feb 2024: The Week in Meep, 18 Feb 2024: Navalny, Insurance Fraud, and Lent Begins

• 10 Aug 2018: Having Fun with Mortality: The Day Nobody Died, the Oldest People Keep Dying, and Don't Drink and Walk