Dr. Shiva Ayyadurai & The Danger Of Data Charlatans

Election Fraud in Michigan? Nope: just how lines work

NOTE: On Nov. 16th, Ayyadurai doubled down on his misleading analyses.

Feel free to watch it if you like — see if you get to the punchline before I do.

The Data

The dataset Ayyadurai works with contains:

  • Precinct-level % of Trump votes among split-ticket voters. These are what he calls “Individual Candidate Voters”, people who did not select a party’s “straight-ticket” option on their ballot.
His slide describing the quantities he’s plotting
Example plot: negative linear correlation between % of Republican voters in a precinct vs. the (% Trump votes from split-ticket voters - % straight-ticket Republican voters).
You… actually wouldn’t expect this at all
Where the X-axis is % straight-ticket Republican votes, and the % of split-ticket Trump votes is some Random Variable. I’m representing the quantity Ayyadurai plots in his video: (the % of split-ticket Trump voters MINUS the % of straight-ticket Republican voters)
When your Y-axis involves a term from your X-axis, you’re gonna have a bad time. Image source: A charlatan who’s trying to pull one over on us

How Ayyadurai Is Misleading You, In Detail

We’re gonna keep this as simple as possible. Let’s say that each precinct has two main populations of people: people who vote with a straight-ticket ballot and people who do a split-ticket vote. Ayyadurai himself frames these as two distinct populations, so I’m rolling with that premise here.

Example from his slides
  • Split-ticket voters, on the other hand, just have some random, non-zero chance of voting for Trump. We will assume this is a flat probability for now (but I show how my conclusions hold for almost any distribution of split-ticket votes, flat or otherwise, later in the article).

So, what’s going on?

Let’s look closely at what we’re doing.

We’re adding a line that’s downward sloping and negative by construction. Obviously this will lead to a downward-sloping final result.

What if the probability of split-ticket votes for Trump isn’t constant?

Let’s present Ayyadurai’s argument in the strongest light. Here’s a common objection from respondents: why do we assume a flat probability of split-ticket vote percentages? What if the probability of split-ticket votes for Trump is linearly correlated with the % of straight-ticket Republican voters in a precinct?

Perfectly correlated, by definition
This gets even more interesting — but stil downsloping — when the correlation is NON-linear.
The actual slope of correlation is 0.6, so I was being pretty generous with my assumption of 0.9 earlier.

Invite to Dr. Ayyadurai

If this is truly what Dr. Ayyadurai meant to do, then it’s essentially just a mathematical trick. A lot of build-up and sleight-of hand for an inevitable and mundane result. It’s kinda like if I asked you to think of a number, add 5, subtract 8, and tell me the result. You shouldn’t be impressed if I could guess your number: all I have to do is add 3 to your result and I’ll know what you started with. In the same vein, Dr. Ayyadurai is conjuring a “suspicious” negatively sloped line from any data that you would likely start with.

Some good challenges from respondents so far:

What about Wayne county? It doesn’t display this negative linear trend.

Simulation when I randomly generate precincts from a normal distribution, where the mean % of Republican voters is low. Looks more like a blob than a line. If you modify the code to also inject noise into precinct sizes, and zoom in on the X-axis, it looks more and more like the Wayne example.
I’d laugh if this weren’t sad.

Focused on experimentation, causal inference, causal discovery, & explainable machine learning. naimkabir.com