In the tumult of the presidential campaign season, Patrick Conway of Mystic came across an item in a newspaper recalling Ben Franklin's observation that a democracy needs an informed citizenry to function properly.

That made Conway, a 78-year-old math and science instructor who taught at Fordham University and Fordham Preparatory School, think about how informed our citizenry was this election year.

It also led him to devise a model for predicting the outcome of the presidential election, based on how informed the electorate may actually be.

Conway, who was a Clinton supporter, started with an unapologetic supposition that the more informed would vote for Clinton and the less informed for Trump.

Pondering the eventual outcome, back in October, Conway came up with a system that pretty closely matched Electoral College predictions by poll analysis guru Nate Silver and Moody's Analytics, which used a formula of complex economic factors.

To find a statistical barometer for how informed the electorate may be, Conway turned to the percentage of people in a given state who have a bachelor's degree.

He then took a list of the states, ranked by percentage of degrees, and split it in half, giving Clinton the most educated and Trump the least educated states.

He then calculated the number of Electoral College votes in each candidate's states.

This led him to a forecast of Clinton winning with 302 electoral votes, exactly the same as Silver's much publicized and, we now know, inaccurate forecast. Moody's was even further off the mark, with 332 electoral votes for Clinton.

In the end, of course, Trump won with 290 to Clinton's 232.

We've heard Silver, pollsters and the nation's sprawling punditariat try to explain away the staggering disconnect between the forecasts and the reality of Trump vs. Clinton.

I've tried to tune much of this out.

We don't have a national reporting staff or polls here — those people have had a lot of soul searching to do — but my gut sure got it wrong, too.

Conway, with the assurance of a scientist, also acknowledged that he got it wrong with his prediction based on an assessment of how informed voters are.

After he gave me a patient walk through of his calculations while we sat at his kitchen table, the careful teacher sent me away with his notes, filled with precise numbers and calculations.

I didn't realize until I had left him that one of the pencil scribblings on the last page, the election results, also included the grade he gave: a little F circled in the top corner.

I am not sure the grade was for himself or voters.

Still, Conway, while botching the forecast, did get it right that the preponderance of Trump-won states were those with lower percentages of residents with college degrees.

Indeed, of the 25 states originally part of Conway's Trumpland, the president-elect carried them all except Nevada and New Mexico. Clinton, though, lost seven of her 25.

But Clinton's losses in the 25 states assigned to her in Conway's formula were predominant in those with lower percentages of bachelor degrees.

Her wins were at the other end of the education achievement spectrum, from Hawaii, where 29.6 percent have a bachelor's degree, to Massachusetts, with 38.2 percent, and the District of Columbia, with 48.5 percent.

Trump dominated in states with much smaller percentages, like West Virgina, at 17.3 percent, and Arkansas, with 18.9 percent.

I wonder what Ben Franklin would think? Was it an informed electorate?

This is the opinion of David Collins.

d.collins@theday.com