If you want to know how Donald Trump can win Pennsylvania in November, I can tell you just by looking at the numbers.

Not by opinion poll numbers, which are ephemeral and sometimes flawed.

Not by speculating about how the campaign will unfold; no one can predict that.

I can do it by looking at numbers that tell how Republican and Democratic candidates have performed in the key media markets in Pennsylvania over the last 25 years. Those numbers reveal patterns in voter behavior and, just as importantly, how those patterns have changed.

Donald Trump can win Pennsylvania if he does four simple things:

1. Keep Hillary Clinton’s numbers down in Philadelphia. No one expects him to win the city, but if he can get 25 percent to 30 percent of the vote, it will help him enormously. His objective will be to cut the margin and, if possible, tamp down turnout.

2. Do well in the Philadelphia suburbs and the Lehigh Valley. By “well,” I mean win at least two of the counties – Chester and Berks are the most likely candidates – and keep it as close as possible in the other counties, losing to Clinton by 4 to 6 points in each. The votes he gets in these counties will offset the margin Clinton will build up in the city.

3. Pile on the Trump vote in central Pennsylvania. In the past, Republican candidates have won 70 percent of the vote in this populous conservative heartland. There are nearly 1 million voters in the 14 counties that comprise the Harrisburg/Lancaster/Lebanon media market.

4. Win big in western Pennsylvania. Trump likely will not win Allegheny County, which includes Pittsburgh, but the Republican Party is resurgent in the former coal and steel areas that surround the city. In 2012, Mitt Romney, whom Trump labeled a lousy candidate, won 11 of the 12 counties in the Pittsburgh media market (he lost Allegheny County) and 11 of 12 counties in the Johnstown market (he lost to President Obama by a narrow margin in Centre County).

So, there it is. Congratulations, Mr. Trump: Pennsylvania and its 20 electoral votes are yours.

Now, what are the odds of Trump achieving all of these goals? On a scale of one to 10, I would give him a two.

Let’s take them one by one.

1. Keep Clinton’s numbers down in Philadelphia. Good luck with that. The last Republican to get over 30 percent of the vote in Philadelphia was George H.W. Bush in 1992, when he won a 34 percent share of the vote to Bill Clinton’s 66 percent. That Republican share has gone down in every election since. Mitt Romney got 15 percent in 2012.

Two factors are at work: the demise of the Republican Party in Philadelphia and the rise of black voter participation. Generally, Philadelphians are very Democratic, but the city’s black voters are fanatically so. They gave Barack Obama 99 percent of their vote in 2008 and 2012. Some have argued that Obama being the first black president had something to do with that, but black voters also gave 98 percent of their vote to Al Gore and John Kerry, both white Democrats.

An argument could also be made that black voters will have less enthusiasm for a white woman who registers low on the charisma scale, translating into lower turnout. Before assuming this, please recall that Trump was a leader in the “birther” movement in 2012, which disputed Obama’s parentage and citizenship. I am sure the city’s black voters will be reminded of that before the election and will act accordingly.

2. Do well in the suburbs. Here, Trump runs into a situation not of his own making. The big trend in the last 25 years is the transition of the Philly suburbs from Republican to Democratic. Republican candidates up until the 1990s could count on the suburbs to cut or eliminate the Democrats’ historical edge in Philadelphia. 

The last Republican to do that was George H.W. Bush in his 1988 race against Michael Dukakis – 28 long years ago.

In 2012, Romney did win Chester and Berks Counties (by thin margins), but Obama won the other suburban counties and the Lehigh Valley by a margin of about 140,000 votes.

Add on Obama’s 490,000-vote margin in Philadelphia, and he ultimately left this media market with a 632,000-vote lead over Romney. 

That made Obama unstoppable. Although he did lose in central Pennsylvania and Romney did very well in western Pennsylvania, Republican gains in other regions of the state were simply not enough. Romney ended up losing the commonwealth to the president by a margin of 309,000 votes.

Now, let’s come back to here and now. Do you think Trump can outperform Romney’s 2012 showing in the suburban counties outside Philadelphia? To be sure, there are angry voters here, but many of them are women who are angry at Trump for denigrating other women. How he does in the Philly suburbs and the Lehigh Valley will decide the election in November. I doubt Trump can do better than Romney did in these southeastern counties – and he could easily do worse.

3. Pile on the Trump vote in central Pennsylvania. This area still is strongly Republican, but the days of a Republican getting a 70 percent margin have long passed. Younger, more liberal voters have moved into the new suburbs in York, Dauphin, Cumberland and Lancaster counties. The presence of black and Latino voters is stronger. The upshot is that while Republicans still win, the margins now average 60-40 in this media market.

The big question I have about Trump here is his standing among white, conservative, middle-class Protestant voters. Will they embrace a hedonistic New Yorker whose style is more Augustus Caesar than Norman Rockwell?

In this area of the state, Republicans who cannot vote for their party’s candidate usually don’t vote Democrat instead. They just skip the presidential race and focus on the down-ballot elections.

4. Win big in western Pennsylvania. This should be a Trump stronghold, where there are many voters who embrace his combative America First message. This has always been the angry zone of Pennsylvania, and voters here finally have someone who matches their dispositions.

The fly in the ointment is Allegheny County, which includes Pittsburgh and remains resolutely Democratic. There are nearly as many voters in this county (925,000) than there are in the other 11 counties that constitute the media market (950,000).

In 2012, Romney won by as much as 68 percent in the surrounding counties, but got only 46 percent in Allegheny County. The pro-Obama volume of the vote in Allegheny nearly offset the pro-Romney vote in the surrounding counties. The result: Romney won this market by just 38,000 votes.

Can Trump blast through the trends and win Pennsylvania by keeping Clinton’s margin down in the city, by trouncing her in the suburbs, by winning central Pennsylvania by 70 percent, and then administering the coup de grâce to Clinton with a huge showing in the west?

Trump’s supporters are encouraged by his showing in the April GOP primary, where he obliterated his opposition, coming in 35 points ahead of Ted Cruz, who ran second. That showing demonstrated the Trump appeal, but beware of making too much of it.

The Republican primary drew 1.5 million voters. Since the total vote in the general election is likely to be close to 5.5 million, there are an additional 4 million voters yet to be heard from. They will get their say on Nov. 8.