Begging the Question

Stop me if you've heard this one: "Our user engagement metrics dropped significantly last quarter, which begs the question: Is our new UI design driving users away?"

Deep breaths. Deep, calming breaths.

No, Karen from Product Management, that doesn't "beg" any questions. It raises questions. It prompts questions. It might even scream questions while running naked through the streets of Silicon Valley. But it does not beg the question.

To those of you tossing around "beg the question" like confetti at a New Year's party, what exactly do you think it means? Are you saying it because you've heard it in meetings and thought, "Well, that sounds smart; I should sprinkle that into my vocabulary"? Much like using "irregardless" or "literally" when you mean "figuratively," misusing "beg the question" is a one-way ticket to the Linguistic Hall of Shame.

A Brief Stroll Through History

Begging the question goes back at least as far as Aristotle. In his writings of logic and debate, he addressed the idea in debate when one simply asks one’s opponent to just accept your initial point. There’s a question at hand, and you beg for your take to be accepted without benefit of argument.

Aristotle’s Greek was translated into Latin as “petitio principii,” which might best be translated as something along the lines of “petitioning the starting point.” It's like saying "I'm right because I'm right." Which, coincidentally, is how most of my arguments with my teenager end.

Fast forward from the Romans to the 16th century, and some well-meaning but linguistically challenged scholars translated petitio principii as "begging the question." And like a badly trained machine learning model, we've been perpetuating that error ever since.

Examples That Make You Cringe

Begging the question might be a statement like, “Churn was high this past quarter because so many customers canceled their accounts.”

Ah, I see. Very wise.

Moliere illustrated this beautifully with one character asking why opium causes sleep, and another character answering “Because of its soporific power.”

In 1968, Chicago mayor Richard Daley explained why Hubert Humphrey had lost the election… because he didn’t get enough votes.

As data scientists, we're particularly vulnerable to question-begging, though usually in more subtle ways than declaring "This model is accurate because it makes accurate predictions." (Though I've seen that too. Looking at you, Steve from Marketing Analytics.)

Or imagine that customer satisfaction scores are missing for 30% of users. And we observe that customers who respond tend to have higher engagement metrics. A data analyst imputes the missing satisfaction scores based on engagement. Later, the analyst - or even a different analyst - conducts and analysis and concludes that higher engagement leads to higher satisfaction

Face, meet palm.

The Perils of Assumptions

Assumptions are the hidden trapdoors of data analysis. They lurk beneath the surface, waiting to plunge you into the abyss of faulty conclusions. It's like building a house on quicksand and then wondering why it's sinking.

In the world of Bayesian statistics, priors are our initial beliefs before seeing the data. But if our priors are flawed or based on circular reasoning, our conclusions will be as shaky as a Jenga tower in an earthquake.

This isn't just pedantry (though I excel at that too). In the world of data science and analytics, circular reasoning can lead to some spectacularly wrong conclusions:

• "Our predictive model must be good because it confirms our initial assumptions!"
• "The algorithm is unbiased because we trained it on historical decisions we assumed were unbiased!"
• "The clustering makes sense because it groups similar things together... which we defined as things that cluster together!"

It's like using a mirror to check if your eyes are closed. Technically you're doing something, but you're not getting any new information.

So, What's the Solution?

First, let's all agree to use "beg the question" correctly. If you're not sure, opt for "raises the question"—it's a safe bet and won't earn you the side-eye from your more linguistically advanced colleagues (like me).

Second, in data science and beyond, be vigilant about your assumptions. Question them relentlessly. Are you inadvertently setting up a self-fulfilling prophecy? Are you confirming what you already believe rather than uncovering new insights?

Look for independent validation. Be especially suspicious when results perfectly match expectations.

And have I ever mentioned that correlation isn’t the same thing as causation? Just checking.

The Deviltry of Lexicographers

Now, I know what some of you are thinking: "Language evolves! Common usage changes meaning!"

To which I say: Not on my watch. Let this be my Thermopylae. I will stand here with my Spartan army of logical purists, defending the proper usage of "begging the question" against the invading hordes of linguistic drift.

What's that about Thermopylae? Oh, right... maybe not the best historical example. But regardless, how can common usage have changed the meaning of “beg the question” when the original meaning is still the way the phrase is supposed to be used?

Questions can be raised, asked, prompted, or even shouted from the rooftops. But unless you're engaging in some sweet, sweet circular reasoning, they're not being begged. And though I won’t correct you when I hear you say it, know that I’m judging you. And so is Aristotle. And trust me, you don’t want to be on his bad side.

Talk to Me

Have you encountered any cringe-worthy misuses of "beg the question" or other phrases that make your inner grammar nerd weep? Or perhaps you've fallen into a logical fallacy in your own work and lived to tell the tale? Share your stories in the comments.