Nearly ten years ago, on November 9, 2016, the morning after Donald Trump shocked America and much of the world by being elected President of the United States, I wrote a piece titled “Pollsters not wrong, they merely failed to ask the right question.”

My argument wasn't that pollsters couldn't count. It was that they were measuring the wrong thing. Ask people, “Who are you going to vote for?” and they'll give you an answer. Properly sample enough of them, weight the responses appropriately, and you'll get a number. But that number doesn't necessarily tell you what's going on underneath.

In 2016, I wondered whether conventional polling was missing an underlying sentiment toward Trump - people who wanted change, people reluctant to admit their support publicly, and people simply saying, damn the torpedoes, let's try something different.

I ended that piece wondering whether America had elected what I rather impolitely described as “Howard Stern, the buffoon,” or whether Donald Trump (like Stern) possessed another side - intelligent, insightful, reasonable and considerably less buffoonish. I finished with three words: Time will tell.

Well, time told.

Ten years later, I'm less interested in telling you what it told us than in knowing what Americans think it told them. And once again, I think pollsters may be asking the wrong question.

THE MEASUREMENT PROBLEM

With all due respect to an industry that was my career beginning with Statistics Canada in 1976, most pollsters still don't know how to bend the edges of the box when it comes to sampling and questionnaire design.

The first criterion is knowing precisely what you wish to measure; the second is knowing how to reach the appropriate sample; and the third is knowing what questions to ask - or, increasingly, what combination of quantitative and qualitative sentiments to record without introducing bias. By the way, in the right hands, AI is good at doing this. More on that another day.

Simply asking people how they intend to vote is old school. It's useful and sometimes highly predictive, but it isn't necessarily very informative. The answer to “Who will you vote for?” can be two or three layers removed from the attitudes, experiences, emotions and expectations that determine it.

This is why qualitative research - focus groups, in-depth interviews and simple human conversation - can be so important before quantitative measurement. It helps you discover what the hell you're trying to measure. Then comes the quantitative phase. After the numbers come in, more qualitative research can answer the most interesting question of all: Why?

So what does this have to do with the current political climate in America? More importantly, what might it tell us about the path Americans - and inevitably much of the rest of the world - are on for the next two, five or ten years?

America has experienced - or perhaps been forced through - a tectonic shift, politically and geopolitically, at home and abroad. The purpose of this piece isn't to argue whether that shift has been good or bad. We've had ten years of that, producing plenty of angst, anger and partisanship but remarkably little useful insight into where America may be heading.

For this exercise, we already have a pretty good idea how Democrats feel about Donald Trump. Pretty much all of them. Measure voter intention, enthusiasm, engagement and likelihood of showing up and you'll learn much of what you need to know. Republicans, on the other hand, need a little dissecting. To say the least.

Let's set aside the labels Republican and Democrat for a moment. Using the language you might hear at a family BBQ, on a golf course or around a bar - anywhere people talk when there isn't a pollster standing in front of them - imagine several broad groups within the American electorate.

BUCKET ONE

I voted for Trump all three times and I'd do it again in a heartbeat. Hell, I hope he finds a way to run for a third term so I can vote for him again.

These aren't simply Republican voters. They are Trump voters, and the distinction matters.

BUCKET TWO

I may or may not have voted for Trump in the past, but what he started needs to continue. Looking ahead, people such as JD Vance, Pete Hegseth or Marco Rubio are the guys I have confidence in going forward.

This may be the most important bucket. These aren't necessarily Trump loyalists; they're potentially Trumpism loyalists. If this group is large and growing, Trump has created something more consequential than an election victory: a movement capable of surviving him.

BUCKET THREE

Okay, that didn't turn out quite the way I expected. If I had to do it over again, I might change my vote. Either way, it's probably time for something different.

Conventional approval polling can easily misunderstand this group. Disappointment doesn't necessarily mean conversion. Someone can be finished with Trump without becoming a Democrat; they can remain conservative, become politically homeless, vote independent or stay home.

BUCKET FOUR

That was an unmitigated disaster and America and the world will be paying the price for years to come. It's time to give the Democrats another shot.

These people have crossed the bridge. Whatever they once thought about Trump, Republicans or the political experiment of the past decade, they've concluded that the answer lies in the opposite direction.

THEY ARE HYPOTHESES, NOT SURVEY QUESTIONS

Make no mistake, you are unlikely to find my wording on a survey instrument anytime soon. If you saw those statements verbatim on a professional questionnaire, there would be enough industry criticism to choke a horse. They aren't survey questions. They are hypotheses.

They are the underlying sentiments I would want research to identify, refine and ultimately measure. That's where the methodological creativity comes in. Don't ask, “Are you a disappointed Trump voter?” or “Do you regret voting for Donald Trump?” or “Are you loyal to Trump or to MAGA?” Those questions practically beg respondents to tell you how they want to see themselves.

Instead, measure the components:

Did the last ten years turn out better or worse than you expected?

Would you vote for Trump again - or for a candidate he endorsed?

Would you prefer a Republican who continues Trump's policies but has a very different personality?

Should the Republican Party continue in its present direction or change course?

Could you imagine voting for a Democrat?

Could you imagine not voting at all?

Is America stronger or weaker than it was ten years ago?

Which changes would you keep, which would you reverse, and related to this, how certain are you?

Don't ask respondents which bucket they belong in. Let their answers tell you.

Use qualitative research first to discover the language and motivations conventional questionnaires may be missing. Build a quantitative instrument around what you learn, segment people by patterns across several questions rather than one convenient headline number, then return to qualitative research to find out what the numbers mean.

And most importantly: measure it again.

Six months later.

Then six months after that.

MOVEMENT MATTERS MORE THAN THE SNAPSHOT

The size of the buckets matters, but the movement between them matters even more. If Bucket One remains enormous, Trump's personal political power survives. If people move from One into Two, something more historically significant may be happening: Trumpism is successfully separating itself from Donald Trump.

If Two begins moving into Three, the movement itself may be weakening. If Three grows but doesn't move into Four, America could be heading toward political fragmentation rather than a Democratic realignment. And if Three begins flowing heavily into Four, the political pendulum hasn't merely stopped. It's swinging.

That's what I'd want to measure - not another snapshot showing Republicans at 46%, Democrats at 47% and everybody arguing about the margin of error. Conventional voting-intention polling is pretty good at telling us where Americans stand today. What interests me is where they're prepared to move tomorrow.

THE QUESTION THAT MATTERS

Ten years ago I argued that pollsters weren't necessarily getting the answers wrong. They were asking the wrong question. Ten years later, I think we are in danger of doing it again.

The most consequential political question in America today isn't “Do you approve of Donald Trump?” It isn't even “Who would you vote for if an election were held tomorrow?” The question is:

After ten years of Donald Trump, what do Americans want America to become next?

Figure out how to measure that accurately - without telling respondents what answer you're looking for - and then measure how those sentiments change over time.

The pollster with the balls and methodological creativity to do that won't merely have a better chance of predicting the next election. He or she may be able to tell us where America is going for the next two, five and ten years.

Again, time will tell.