Respectfully submitted, this is meaningless without a statistical baseline to compare it to — especially when the timeline is over 4 decades long. If you let any timeline play out long enough, statistically unlikely things happen: & you haven’t even gone the distance yet of showing whether or not this is statistically unlikely. How many unusual deaths were there during the Reagan years, or the Bush years? (the latter would be a better comparison given father and son presidencies, tho you might still have to extrapolate over 40 yrs) How many unusual deaths is an average person connected to in 40 yrs? (bad comparison since Clintons aren’t “average,” but helps anchor some kind of baseline) How many unusual deaths is an average “connector” (in the Malcolm Gladwell sense — someone who explicitly knows a lot of people) close to in a similar time frame?
Without the rigor of that framework we are much more likely to draw on inherent psychological biases to evaluate the probability — for example, my gut intuition would lead me to say, “gee, 91 people even in 40 years feels like a lot because compared to my own life I’ve only experienced the deaths of a fraction of close individuals” — but my intuition is leading me astray here, both because i’m comparing my life from birth to the Clintons’ span of actually working in politics, and because even though i’m a “connector” in my own career my Rolodex is still much smaller & its members less “notable” in the public eye. It’s not a good comparison — but it’s the only one i am able to make because no other comparison set has been provided.
It is reasonable to look at this data and have it trigger our suspicion, but we have to be very careful about drawing out false meaning from it — because it can so easily discredit our other valid arguments. It is easy to do, especially when we have a hypothesis we’re trying to prove. It is entirely possible that this is a valid argument also (that the suspicious deaths are unusual and worth heightened scrutiny), but there’s not enough data being shown here to support it.