The media is attempting to capture the public’s attention with poll numbers in the hope that Trump partisans will become despondent about a looming loss, and then not take the trouble to vote.

Four years ago the conventional wisdom was that Hillary Clinton’s massive organizational advantages when combined with incumbency of the Democrat Obama Administration, validated the near-universal view of pollsters that Clinton was going to cruise to an easy victory.

But the combination of “conservatism,” “nationalism”, and “populism” that Trump combined together, and then took directly to the electorate over the heads of media produced an “electorate” remarkably different from the one predicted by the pollsters in their models. But polling is polling; it’s an art, not a science. One thing not factored into the model changes from 2016 to 2020 is how changes in party registration alter the composition of the electorate that shows up at the polls.

Pennsylvania reflects a bad series of numbers for Joe Biden in this regard, and it explains why Pres. Obama is headed to Philadephia this week.

As a general proposition, having more new voters register for your party than the opposition party is a good thing because only a very small fraction of new party registrants are likely to vote for the candidate of the other party.  Party switchers are great, but they are generally less plentiful.  New voter registrants tend to vote as a higher percentage than other registered voters.  This is why both major parties work so hard at voter registration.

So, what does the recent history of voter registration in Pennsylvania reflect?  Let’s look at the numbers of the last four Presidential election outcomes in Pennsylvania as our backdrop — both the percentages and the raw vote totals:

2004 — Kerry 2.938 million (50.9%); Bush 2.794 million (48.4%):  +144,000 for Democrats

2008 — Obama 3.276 million (54.5%); McCain 2.656 million (44.1%):  +620,000 for Democrats

2012 — Obama 2.990 million (52.0%); Romney 2.680 (46.6%):  +310,000 for Democrats

2016 — Clinton 2.926 million (47.5%); Trump 2.970 (48.2%):  -44,000 for Democrats

In the last two election cycles, the Democrat gross vote total declined 350,000 votes.

In the same two election cycles, the GOP gross vote total climbed 314,000.

This wasn’t just a matter of who the candidates were, however; it was also a matter of who the voters were.  And that is where the voter registration numbers become important.

In 2008, there were 4.480 million registered Democrat voters in Pennsylvania, and there were only 3.243 million registered GOP voters.

That was an advantage of 1,230,000 more registered voters for Democrats.  But even with the wildly popular Barack Obama making his historic first run, he captured a vote total equivalent to only 73% of all registered Democrats.

The less-than-wildly-popular John McCain, on the other hand, captured a vote total of 82% of the number of registered Republicans.

Obviously, there were cross-overs — GOP registered voters cast their vote for Obama, and Democrat registered voters cast their votes for McCain.  But just as a function of the raw relationship between the number of registered voters for each party, and the number of votes cast for each party’s candidate, Obama underperformed McCain while Obama was receiving more votes than any candidate ever in Pennsylvania.

The same calculation got worse for Democrats in both 2012 and 2016.  In both election cycles, the number of votes for the Democrat as a percentage of Democrat registered voters went down, while the same number for the GOP went up in both years.

In 2012, Romney’s vote total of 2.680 million represented 86% of the number of GOP registered voters.

In 2016, Trump’s vote total of 2.970 million represented 90% of the number of GOP registered voters.

Pres. Obama beat McCain when the numbers were 73% v. 82%, and Romney when the numbers were 70% v. 86% — with his margin of victory cut in half from 2008 to 2012.

Hillary Clinton lost when the numbers went to 69% v. 90%, and the respective vote total differential went from a comfortable 310,00 vote win for Obama to a narrow 44,000 vote loss for her.

The mantra is that Biden only needs to do a little better than Clinton did in a few Pennsylvania counties where Democrats dominate, and that will be enough to flip the state back narrowly into the Democrat column on Nov. 3.  That is a wild oversimplification that ignores what has been happening in Pennsylvania the past 20 years — and more importantly, ignores what has been happening in Pennsylvania the last four years, as my colleague and RedState polling/voter data expert Scott Hounsell explained last month.

Here are the final registration numbers for 2020, and the change from the 2008 “high water” mark from Obama’s first run.

Democrats:  4.169 million  (-311,000 since 2008)

Republicans:  3.452 million (+209,000 since 2008).

If you re-run the 2016 election with these registration numbers and the votes/registered voter percentage from 2016 (Clinton 69% v. Trump 90%), you get the following outcome:

Biden — 2.877 million

Trump — 3,107 million

Trump wins by 230,000 votes, and the margin is 51.9% to 48.1%.

Pres. Obama’s high water mark in the votes/voter registration percentage was 73%.  Let’s give that to Biden while keeping Trump at 90% which he got in 2016.

Biden —  3.043 million votes

Trump — 3.107 million votes

Trump wins that contest by 64,000 votes.  Biden does as well as Obama’s historic run in 2008, and Trump does the same as 2016, and because of registration changes, Trump wins.

If you use the “average” of the past 3 elections for each — Biden would be 70.67% and Trump would be 86% — the results would be.

Biden — 2.946 million votes

Trump — 2.968 million votes

Biden still loses, but by a razor-thin 22,000 votes.

So to get Biden a win, he needs to do better than all but Obama’s historic win in 2008, and Trump has to do significantly worse than he did in 2016.

This registration pattern, and relationship between votes received and voter registration, has moved steadily against Democrats going back to 2008, and it continued from 2016 to 2020.  Democrats still enjoy a sizeable advantage in voter registration numbers, but a significant number of registered Democrats have consistently voted for GOP Presidential candidates for two decades.  And the registration advantage has declined from 1.23 million in 2008 to only 717,000 now.

The most recent 6 polls for Pennsylvania on RealClearPolitics have Biden winning with an average vote share of 49.2% to Trump’s 44.8.

The last four Presidential elections have had a combined vote total for all candidates of between approximately 5.8 and 6.1 million votes.  If we use the average of approximately 5.95 million votes, then raw vote totals of Biden and Trump based upon the polling average would be approximately:

Biden — 2.927 million

Trump — 2.665 million

But let’s juice the total voter turnout to account for high interest — let’s place it above 2016 by the same amount 2008 was above 2004: +200,o00 overall.  So let’s set 2020 at 6.35 million.

Biden — 3.214 million

Trump — 2.845 million

So in this scenario, Joe Biden would receive more raw votes than Obama did in 2008, and 300,000 more votes than Clinton did four years ago.

But the real fantasy is when you compare those raw vote totals based on polling to the registered voter numbers, and determine the percentages such an outcome would reflect in 2020:

Biden —  77%

Trump — 82%

The AVERAGE for votes/registered voters for Democrats in the last 4 elections is 71%, and the best Obama did was 73%.  Yet the polls say Biden will do 77%???  That 4% difference between Obama’s best performance and what the polls say Biden will do, represents 168,000 votes for Biden than Obama got in his historical first campaign.

The AVERAGE for votes/registered voters for the GOP in the last 4 elections is 84.5 — including 80% by Bush 43 and 82% by McCain.  Getting only 82% as the polling claims would represent a HUGE Collapse of support for Trump among GOP voters after his number was 90% in 2016.

Heck, Romney’s number was 86% — and Trump is going to underperform Romney with GOP Voters in Pennsylvania?

An 8% decline in votes when measured against the GOP’s 2020 registered voter numbers means 276,000 fewer votes than 2016 from GOP voters.  That is a 9.3% decline in RAW votes compared to the number of votes Trump received in 2016.

A quantitative analyst at Morgan Stanley has done a study of the relationship between changes in party registration and vote outcomes.  Zero Hedge covered it in this article.  Based on a similar evaluation, he sees almost every battleground state breaking for Trump in the end.

Now you know why Barack Obama is going to Philadephia.

And why the polls being published about the state of the race there are worse than meaningless — they are propaganda intended to suppress the GOP vote.