battlefield pennsylvania

It’s been a windy hop through the 2024 election season. We saw most recently in September Kamala leading by 1-2 percentage points, in the average of national polls.  Since the August DNC, I had suspected that this would henceforth be a locked-in margin, but for any extreme unforced errors by her [to be sure Trump made his own earlier].  Equating to a 65% or so chance of winning the popular vote.  However, the Electoral College [the vote that matters] has now revealed itself to be particular.  You instead have to look and weigh tabular polling records, state-by-state, and relative [more accurately square rooted] to size.  Even among swing states, some [e.g., Blue Wall states] were easier to gauge than others.  Others challenged such as Wisconsin; perhaps Arizona.  In our examination we see that the main pathway for both candidates still reduces to mostly just one state: Pennsylvania.  The probability data coming from there has deviated from the national averages in multiple large ways [and in ways only unique to that location], accessing only advanced math to appreciate, and ultimately proved to benefit Trump by a reasonably stable amount in general.  This “surprise” should not have happened in many traditional setups within a single country.  And this is not a normal election season. Why are we here, and what are the best things candidates should do at this point?  Given its proximity to our Princeton bubble, Mira👶 and I decided to venture there to better understand the modern nuances of this statistical battleground.  Our article explains what we’ve learned from the polling math, and away from institutional news flow. Also what it reveals about the state of some neighbors plus their ingestion of information.

More on Pennsylvania

Looking at the past several regional and national elections in this state it made sense to focus on its eastern third, as the frontier most inclined to switching.  There is a multi-hour driving arc shown below [in purple] that runs through west of Allentown, straight down from Scranton and away from Philadelphia.  That’s where I wanted to focus my efforts for this weekend.

But can’t other states instead of Pennsylvania be the ones that matter?  Of course, but there are also some issues with just this line of thinking.  First, in our probability theory one must center their exam on the most likely 1-2 states, knowing full well that the actual results will almost certainly lead to surprises not just outside of Pennsylvania – but in a state or two completely different from those currently hyped in the news.  Second, the next states in line to matter [after Pennsylvania] also have fewer electoral votes. Exponentially trickier to analyze as part of a larger combinatorics problem, and certainly not on a budget [which everyone is on due to limited resources and a mission to best target mindshare for this election cycle]. 

Versus Pennsylvania’s 19 votes: Arizona has 11 votes; Wisconsin has 10 votes.  Interestingly too they are geographically spread very far from one another.  They also have little, culturally, in common.  But their common statistics this time is another matter.

Staying with Pennsylvania though, while integrating cross insights from the other major geographies as well, we see that the “top” issue that voters self-declared revolves around the economy.  But not in a plain way.  First take a look at national polls, and then we’ll peer into these anomalies within states.  

 

National polls - voting intent

49% Kamala [+2%]

47% Trump

 

National polls – “top” issue trust

46% Kamala

48% Trump [+2%]

 

Now indeed for those voters [which are not that many who singularly focus on it] Kamala has the higher edge nationally on her strengths [eg on abortion she’s 54-55% nationally but 53% in Pennsylvania], however as we’ll provide below this only tells a small part of a multiplex story that we need to better realize.  For now, we’ll also state the next two “top” issues are nearly equally significant.  Those next two issues are -first- the economy [which is open to interpretation from one person to the next in terms of being anything from finding jobs, to asset stability, to inflation].  And -second- immigration. 

Now both of these two latter issues Trump would have a double-digit edge on.  And we have no detailed models -connecting to any one state poll- on how these issues relate to one another [very few voters can be assumed to have just one of these “top” issues].  So in the interconnected dimensions that matter, the polling can’t allow a prejudiced person to wrestle with the idea that people can have near infinite mix-and-match between conflicting platforms.]  This notwithstanding, all three collectively of these “top” classes still only stand for just over half of the population’s “top” issues! 

Said differently, candidates are fighting for a small percentage edge at most, on a frighteningly small sample size, and running off data that is half biased!  2024 and we are well within the world of big data and AI; yet we don’t have a solid read on how a voter might be influenced any time before election day.  Further, where these top issues stop, we have voter considerations that outright devolve into randomness afterwards.  As an example, the next in-line issues of significance: such as democracy, or dislike for other candidate.

How small of a sample are we talking about?  Take young women, where large state polls have a narrow sample ~60 women surveyed between ages 18 and 29.  So that works out to a sample of just 11 per every 2-age group [eg 18 and 19 year olds combined], and further, only 1 per “top” issue for same every 2-ages!  Is this a joke? I run 2,000 sample social media polls nonstop, with cells sample sizes that exceed this.  A single cached respondent can swing an otherwise 50/50 split in a 20 person cell, to be 11/9, or a 55% versus 45% [false +10% or more than enough every month or so to randomly flip directions on who leads all the way up to the national winner in a tight year!] 

Returning to the young women example, for the area of Pennsylvania I travelled to now, this polling sample size you see from major news outlets reduces to less than 10 people unweighted.  Contrast that with my personally this weekend physically confronting and talking to 35 people, and already post-weighted.  Plus the ability to explore where the conversation leads and ask for explanations. 

Other issues are we are undersampling across the board on key issues and sub-demographics, there is also a bias in who is sampled from pollsters to pollster, we also don’t really know how said voter will actually vote on that issue nor how they weigh it against other issues.  The polling is also often staler [many weeks in-between!] for battlegrounds even relative to their high-frequency importance and need for precise messaging.  And last, as we’ll discuss in greater detail below, we don’t have a tight correlation, let alone causal, understanding of whether an issue matters anyway [note in national aggregates Trump leads on issues whereas in those same polls Kamala still leads on voting intent.]  We also note most pollsters don’t have nor provide the complete and comparable cross-sectional demographics, adjoined with their polling.

But let’s start to consider these bivariate only cross-sectional polling data.  First we eluded the small data, particularly on state polls. We don’t have reliable large data on the characteristic intersections that are helpful for voter analysis and ultimately messaging.  For example we see across the country’s crosstabs that abortion really doesn’t matter equally to all young voters, or even all female voters, and as we see in Pennsylvania the micro sample of young women voters are not the same from state to state.  There is no one unifying model about the risk of the differences on an important yet heterogenous issue; perhaps there should be, but our state of technology has not yet caught up.

Let’s work out an example.  Nearly 7% of Pennsylvania voters are expected to be young women, versus roughly 6% nationally.  Nationally their top issue remain economy but in Pennsylvania it is abortion.  While here I noticed a disproportionate fraction of the many Kamala ads, seen by anyone in the public from gas stations, were concerning abortion.  Despite trusting Kamala more on abortion [eg approximately 53%] there is an entire portfolio of other Pennsylvanians and other issues that interact in higher importance towards Trump.

The net result is these Pennsylvanian women favor Kamala by a much smaller margin [roughly low 50s% versus low 60s%].  They need to be messaged very individually and thought about in the context of being more complicated than the national analysis.  Be humble about the sample size and mixed signal that national messaging has been flooding these regional women.  For Pennsylvania in particular, they are also more susceptible to changing their mind and voting differently.

 

Pennsylvania versus Wisconsin polls - voting intent

48% versus 48% Kamala

49% versus 48% Trump [+1% in PA, tied in WI]

 

“Top” issue among women

25% abortion in PA [misaligns with rest of PA; not helping Kamala as much]

20% economy in WI [somewhat aligns with rest of WI; helps Kamala greatly]

 

Pennsylvania versus Wisconsin polls – “top” issue trust among women

51% versus 54% Kamala [+2% in PA, +9% in WI]

49% versus 45% Trump

 

Leaving the previously described polling sample errors aside, this is still and actually large compounded, transmission-mix difference between issue priority and voting intent. For micro messaging for a battleground, if Kamala is looking to a banal “top” issue and building a campaign that can bank voters on mostly that. And can also explain half the difference between some swing states and the national averages!  Maybe it’s poor sampling using the crude factors available to pollsters and campaigns, or the next result of messaging by this late stage of the cycle.  We explain below our hypothesis for what’s going on. 

If there are extreme vagaries in who the remaining “abortion centric” voters are in Pennsylvania, then you’ll be making faulty assumptions.   I had conversations with young women through the state that confirmed this.  A mis-message means you had a chance to resonate to someone on the fence, and instead you lost their vote to the other candidate who may not even have had the chance or budget to otherwise engage said person.  Your messaging and dollars essentially 100% transferred over to the other candidate.  And many small misses add up in this case for Pennsylvania and some smaller battlegrounds, and in a way that deviates from the national average.

Recall again the insight that even the exact demographic DNA match-up behaves differently from state-to-state [with Pennsylvania’s 19 electoral votes making it the most important prize to understand].  Imagine messaging a young woman with an ad about abortion, when she is equally struggling with the cultural distaste for how an immigrant may be harassing her at some recent time or have other economic struggles that that weigh more heavily on people here [versus some issues perhaps with people elsewhere].  More people than you imagine are awash with many concerns more complicated than just a single one.  When you admonish someone presuming to know all that the person has been through, you miss the fair mark as they may be someone different from your purposes. 

And when you take relative odds and place the weight of your chips incorrectly, you get suboptimal results.  In particular this is a key to the uniqueness of evolving battleground states, which makes them the trickiest to model from polling and then apply in broad strokes [eg “white women for Kamala” as if they are a universally knowledgeable and relatable monolith]. 

 

Battle fatigue

We also have the exhaustive nature of these surveys, connected to the higher frequency of them, leading to worsening quality of engagement and fatigue.  You can have someone in a subsection of the Pennsylvania [or other] state asked many times [in many forms and levels] who is not even as likely to vote, as others not asked at all [or simply refuse] yet who are more likely to vote.  And the subsections are too large else risk having incomplete cells and corrupt models throughout this process.

Even with war chests totaling billions of dollars and all the polling a modern population can buy, no one can say what’s the relative importance of the #2 or #3 issue among young women in one state versus another.  The singular focus on what’s presumed to be just their rank #1, with all the uncertainty we described around even that, and also unappreciative of the utility of #1 being weaker in battleground states [or in relation to other "top" issues ranked #2, #3, and beyond].

We already explained that we know there is a slight tilt again toward a younger demographic in Pennsylvania, but neither by much nor also a location where the messaging around abortion just didn't resonate as strongly.  There are multiple cultures and work classes crisscrossing even here so one needs to be more careful versus elsewhere. For whatever reason [likely some valid], Trump’s messaging on the combination of the economy and immigration are working well.   And the weight of this information is far greater for its 19 electoral vote contribution, than for its underlying popular vote contribution. It may seem unfair, but that’s another discussion; and for another time.

 

Failed models; is this 2016?

Straight out of copula narratives and surprisingly so that I hadn’t even included election polling as one of the many case study topics in my summer mathematics best-seller.  Focused on primary characteristics and their errors this go around, if you miss the true multisectional information you instead lose the real show.

It’s also extremely important, and an invaluable lesson from our road trip, to always know how to locate and explore one’s blind spots.  I am spending my weekend here against all lay advice from my many friendly followers. What deep introspection are you doing to understand truth's center?  Be vigilant as the dimensions of your blind spots are real, and evolving through life [both yours and of course everyone else’s way too].  The moral of living in an unbiased way also means all of us have to live with greater uncertainty including in how much we understand about one another, let alone where we're headed. 

Far more important than who wins an election is your singular focus to appreciate on your own the miracle of what America’s diversity both offers and represents.  It’s ok to be frustrated regardless of your status in life, and to not always [or even often] have outcomes go a simple way.   Perhaps it is your approach that instead needs adjustment; give up some nights and weekends yourself, and invest now for your remaining time.

In this election, and when dealing with egos and far less quantitative political hacks, it is clear that univariate spectrums have misled, as the notion that all subpopulations must feel similarly and that this was the only lesson we must fight in the aftermath of 2016.  But instead we mentioned in an earlier section that we lack all of the mathematical information about how said distributions vary across the country, how to predict their trends [let alone confidence], and what the cause and effect drivers are this time.  One manufactured table here and there can’t give you much of that. 

The fatality of mistargeting is also more than simply a lost expense, it is a poor reflection of the candidate’s message and a switch in the mind of an undecided, towards the other candidate.  Especially if repeated to small clusters, say on the college campuses we visited which were all different from one another but also perhaps more apolitical that normal.  How could this not have been discovered earlier?  What campaign advisors want to take the immediate career risk of telling a candidate that they have to change strategies by so much yet again?  And yet in this experiment it was never one message or interaction, but rather the cumulative sea of information individuals have through life that figures out if something is working, or not.  And if a candidate issue is excusable, or not. 

Recalling this is an election more about voting against either of these low-favorability candidates, than it is about voting for either of these low-favorability candidates.  While Kamala campaigns to woo people who dislike Trump, Trump is equally campaigning to woo the equal numbers who dislike Kamala.  That’s a very challenging context for either candidate to be in, and unfortunately both have had enough ego and ignorance to think they can defy this undercurrent. 

I am happy to have gone and to reinforce these learnings; they will endure long after this election cycle.  And thankfully there are enough elements of Pennsylvania society that stay above this delicate fray.

 

Missing factors

There were many variables and thoughts that were worth considering, and not covered in the polls.  There are complexities to issues such as the economy that people conflate.  One person in California may see a low national unemployment rate; someone else in Pennsylvania may have had their job lost to some who looked or behaves different [or still wondering if it was due to that reason].  Now both may think the economy is their core issue, but they think about it differently and assess of course their leanings accordingly.  You would be wasting efforts talking down to a person in Pennsylvania using points solely understood to a different person in California [or even general talking points about creating regional jobs]. This is why I wanted to travel here.  So much better in life to ask and see, than simply to guess. 

For example understanding the relative importance of why something matters.  We are guided by physical laws.  Prices rise, societies advance, America leads.  If something happens, it happens regardless of who is in charge, and an issue is important regardless as well.  We haven’t had a President for months [Biden awake?], and yet the world hasn’t turned upside down.  The emphasis on some parts of who is in the White House is moot.  We need to be cautious assuming an issue is always relevant simply as declared, since there may be little one can do about it.  We instead need to know what is it about the issue that is important, and whether it is important enough to be changeable at this time.

Additionally, information needs to be far less biased.  We noted there could be undersampling of Kamala voters in battlegrounds [or perhaps male conservatives in the region].  Or perhaps a larger mix of people who are different in their approach to certain polling types.  My approach to being on the ground was to look for added information [I went into grocery stores and documented prices myself and observed carts, instead of staring from afar at government-manufactured economic data], and talk to dozens of people.  Not merely read the news, or go somewhere and count political lawn flags or signs [which there were so many of from both parties].  It did break my heart to think of unaffordable prices and tough decisions about what one can afford to eat. I had to remind Mira👶 we personally throw around money in the family on simple things that others consider a stretch luxury. 

The economics of this region of Pennsylvania appeared middle of the road: I saw middle class people, and I saw lower class people. Still, I was struck at how beautified and well-functioning many parts of this area of the state was, though to be clear I only drove during daylight but didn't venture to on foot at night, the most frightening neighborhoods.  People lived modestly, people were quiet, and again people were modest.  Even near the sketchier sections of say Lehigh Valley, I saw little in the way of dodgy establishments, such as liquor stores, payday lenders, massage parlors, rotting factories [other than visiting the famed Bethlehem Steel], seedy restaurants, etc.  The valleys has reinvented itself so much, with new highways and industries.  The public schools are middle of the pack, but with a growing commuting demographic those tended to be childless.  There were sporadic complaints about socialism but unclear how much of an impact people here currently have to it.  Prices were inline with the high prices I would see in an upperclass town perhaps.  Stores were not cheap and different stores seemed to cater to different cultural personalities [eg some bookstores that seemed to have conservative friendly books about guns and church and others that showcased rainbows and liberal autobiographies]. 

Prices were more fair if looking at it from my bubble, but clearly a stretch for the incomes of people who could only afford to live here. Still tough to know exactly what else to make of this even as “economy” is almost everyone's “top” issue.  I definitely saw homelessness, though my family didn't. They didn't know how to spot it; since not one person asked me for change [but they did stare.]

If this is what center of the center means, it felt good to see on whole.  People I spoke with didn’t always see themselves or even know how critical their state was or why, just that they were the targets of many annoying politicians swarming their screens -unasked- for months now in ads.  They also didn’t as much of a historic identity, perhaps a reason in this case of the apolitical nature of this population.

One also has to see actual behavior among different groups of people.  I eavesdrop on people’s screens, or pretend to not listen to what other people are talking about.  And instead of asking an explicit question when I do engage, I also ask them for indirect information.  Of course tough to tell based on obvious outward signs or cars driven, and that’s good since as approachable as many people were here they were also more apolitical than I’m used to.

I also don’t just ask if Trump is better.  Why would someone want to tell a stranger that?  Instead  I can play the mark, suggesting I am interested in moving to the area.  So “what are the good schools around here?”, “given women’s concerns would you recommend this state for college?”, ”where is it easy to find a job?”, “where are new families moving to and where can one find a cheaper home than so-and-so area?”, “what international cuisine do you recommend and who works there?”, and “why don’t you keep upscale [or downscale] items here?”  Get information through stealth small talk. 

 

Research setup

It’s always a worry of mine that the surveys are missing something, and usually it's in the relationships among variables.  Not straight to see based on how most people consider statistics.  In this election there are issues that are hard to get at.  People may not want to admit aloud they don’t like immigrants, but yet they 100% vote in a way that [coincidentally?] aligns to that sentiment.  Or as noted many times here people may in fact believe they’d be better off under one candidate but still misbelieve that the other candidate is even better or more “appropriate”.  Or people may not think abortion is important to them, yet ultimately focus on their daughter’s needs when it’s time to enter the booth.  Or people get skittish while waiting in line to vote. 

I don’t envy the stress people here have knowing their vote counts more than most; and that their private compass will have impacts on others far away.  How many ballots are marked not with confidence, but rather with hesitation?  And for an analyst like me, there are too many unknown ways to signal crisscross. 

Most on social media said my coming out here was the dumbest of ideas.  But of course too, why I had to do it.  There are things you see when you travel the world over time.  The world is splendid, but only if you are willing to look, see, and understand it.  Many of my work colleagues would benefit from talking to poor people one on one, instead of talking at them from a distance. And that context has helped me in my engagement with people, because ultimately you will have to get in people’s faces.  Chat it up, and very often building one new connection at least.  And it’s equally important to say none of the statistical analysis here matters if people don’t believe you care about them, and that you are not embarrassed by their life or how they go to that point.  You can’t just tell them you care; they either feel that way about you or they don’t.  You either talk interestedly about their "story" and give individuals your attention, or you don’t.  And I talked to many people this weekend surprised at the chance to open up, laugh through the hardship, follow-up on curiosities, and also learn about statistics and where four more trips around the sun may land them.

But it was important to see countermonoticity from copula narratives in action. It’s difficult to otherwise experience from just theory.  And there are too many issues in America that have lingered too long, and too many false lessons about the meaning of the 2020 election.  We’ll never know for example what the result would have been if the election were instead held on November 2019, instead of November 2020.  And the factors around 2020 aren’t really repeatable for considering other elections, and yet people assume 2020 is a sign the country has completely defeated an “erroneous” 2016 Trump presidency.

We’ll describe below that Kamala needs to have a different message within groups at each battleground state.  Not the same message; different outfit and accent.  She owns the campaign at this point.  Not every idea will work.  Giving free “loans” to entice new voter M, may instead turn off earlier voter F [likely made far worse if voter F is in a battleground state or voter M doesn’t budge].

 

Conclusion

There is a rich amount of information in this part of America, and again particularly for this go around.  Pennsylvanians don’t have to represent a core secular shift in America, but when condensed to a simply binary [or ternary etc] vote they are straying from the national norms [along with some of the smaller battlegrounds] and appear to be align to some political matters that indeed represent the future, but this too could be a false association.  Does America plan to continue serving up pairs of bad candidates?  We have to wait until we get good candidates to re-assess.

Regardless of this election, until something new comes along they are a decent harbinger for now of things to come.  Things are definitely tight.  The state will likely be closer than most in my view.  So it’s hard to predict the election outcomes here, but it seems fairly tight with a faint edge towards Trump.  So unstable however and faint, that I worry about election integrity concerns.  I think it's going to be close and then get ugly in the days post election.  Leaving this aside, there is still valuable advice for both candidates further below to that should follow.  But I know some many not always do so, and eventually a price will be paid.  It's all fair game.  I was there publishing this for a senior campaign advisor in 2016 when Hillary completely lost an election where I identified fake polling in advance of, but she falsely believed and yet Trump was able to better take advantage of. 

Also again hopefully we will get to a community in America where we are voting for candidates instead of against candidates.  Pennsylvania at that time may not be the same battleground it is today, and another state may suddenly rise to that occasion.  Right now we are winding down the clock on this 2024 experiment with some unlearnt lessons. And almost all people now have decided. 

And why are people in Pennsylvania different?  Voting for Trump while most of their neighboring states are voting for Kamala.  How did they get inundated with different messaging over the past couple years?

Technology and social media play a role in how people get non-establishment information.  And in particular independents.  Social media is the largest [and over this year growing even larger] role versus traditional news media.  Some of it is true, some of it is fake.  All like everything else.  And in battlegrounds is highly fragmented in other nontraditional sources as well, and perhaps a reflection of lower appeal.  But we can’t sit on the coasts, pretending this isn’t an element of the future.  The country wants different things beyond what either candidate represents. And the fix will lead to pain en route. 

 

Top news source

22% social media nationally

17% social media in Pennsylvania [tied with Fox and my physical observations is that there's some overlap]

18% social media in Arizona

16% social media in Wisconsin

 

Battlefield Pennsylvania is a story boiled down to a nonchalant population forced yet again to pick between two sub-par candidates, and this is how it plays out for this year.  And it’ll offer surprises the next go around and there is migration patterns both in and out of this location.  Indeed some feel drained about the swarm of unwanted attention; perhaps more than those electrified by it.  It's a turn-off after a while, and that while was months ago!  They'll let their voices be heard, as they see fit, at the voting booth.  And if Trump prevails it'll be tough to know how much of this was economic concerns for some hard workingclass people here, or a tinge of anti-immigrant fears, or simply slightly luckier messaging that resonated a little better at this particular moment.  It’s in some ways easy to have seen a mix of discontent reveal itself through many diverse battleground states all with unique “top” issues per demographic mix.  

I didn't notice much in the way of help wanted signs but I did notice the prices. Kamala may have to contend with a perception to a layperson that a part of this uniquely squares into her time in office.  I didn't meet many people who felt the inflation recently was simply a normal event, but something the current administration did in some way.  I don't think some care what a national TV pundit says to the contrary.  We could all argue some of this is broadly unfair, or maybe not.  Perceptions recede, but this is the state of what I saw and it is possible that some would still be concerned about economics even if inflation hadn't been so alarming.  The current administrations globally have alienated pockets of people throughout the western world, and this is one of the ways it is poking out at this time.  Also recall from some mix of many of these “top” issues, people are saying Trump is the best of the two candidates equipped to deal with current problems. 

There also doesn’t need to be “a reason” why eastern Pennsylvanians care more than elsewhere about “top” issues differently, but the polls and my visit revealed it more strongly in the mix but also slightly finicky.  And that’s certainly something Kamala in their view happened to not be as convincing about in order to dominate most strongly right now. If it wasn’t all here, it could have revealed itself in other ways such as a weaker popular vote poll nationally; statistics is entropically amorphous in that way.  People somewhere are bound to ask the serious question of how will she not continue mistakes she won’t even admit to?  Polls appear to show this line of reasoning looks critically weak and deflecting, at least to enough people to give Trump a very slight edge in total and at this time.

And last in summary, I am concerned that abortion was simply oversold as an issue, but in the end it didn’t matter in combination with other issues and people [especially undecideds] in this region.  Maybe not all young women are thinking this is going to be an issue, maybe rationally or maybe irrationally.  The lifestyles here didn’t seem that unusual from the rest of the country, and perhaps things again may change closer to the election.  But this is not a liberal metropolis and as noted in the article doesn't hold the same passion about this issue that Kamala otherwise commands.  And again, combined with other issues having more influence particularly in a copula-style combination.

 

Advice for Kamala

Unfortunately from an electoral college perspective, she is an underdog [see my math below].  Based on polling I have her in the low 40% range to win the electoral college and that’s giving her the rosiest end of the probability spectrum within my formulae.  It also unbiases any recent critiques [which I originally had] that one may see from betting market odds in the news.  Kamala should outright have different customized messaging again in the battleground, and avoid the need for national messaging now.  And what possible logic led to going on Fox News this week?  Not to be hard since many of the interviews were really good and smart to do.  But nothing was gained from the Fox News appearance, and one this location perhaps saw a little more than others, and an unneeded sign of internal desperation.  It was a combative discourse, one she visibly appeared to regret, and didn’t even include the economy, or even her strength of abortion!  It’s precisely the polar opposite of what campaign risk management would have advised, in every way.  Polling was already weak but yet decidedly higher, prior to the start of the recent interview spate.  Some of those interviews she was also unprepared for, based on the mathematical information here. 

If she does anything now, she needs to provide applied solutions, if she can, to very specific issues people on the fence care about.  Blaming Trump as well has been well-worn this year, without evidence it is working.  The potential blowback isn’t worth the chance it seems.  Perhaps try a different note as this again is hardly the big issue and easily outweighed by many other issues [incidentally Trump too can learn a little from this, though his criticisms more tightly marries her with specific issues, as opposed to only harping on Kamala inside an imaginary void].  We’re well into the statistical final hour and the battleground seems slipping away; and bluntly this crowd may cheer but isn’t going to vote based on a Vogue cover.

probability for team trump
= [45% for PA] otherwise [probability trump winning AZ and WI]
= [45% for PA] and [100%-45%]x[100%-80% for harris winning either AZ or WI or both]
= 45%+55%x20%
= 56%

— salil mehta (@salilstatistics) September 30, 2024

Advice for Trump

I am surprised as most others, that Trump is still in this game, for election season 3.  Apparently this is what America has decided.  Yet pitted against Kamala, here we are.  Commendably he is now well outperforming his wasted 2020 [though underperforming 2016].  His ground game appears better as well, as his eccentric influencers actually resonate with battleground voters, and lubricated with Musk and his platform.  The platform appears to be "get things done" types, and it has worked even though it's not always clear what or how they get things done.  They speak directly to Kamala’s flaws, and hyperactivate fears about say immigrants.  It’s not genius; the peg happens to fit the hole!  For example, very many know there is no mass-scale epidemic of dogs and cats eating across America [not even in Springfield!], but his fiction resonates better than her fiction.  Not everywhere, but happens to be in some places.  And again if everyone in the US had their addresses randomized after the last election, by this point the popular vote could be a dead-heat instead. For now selling fear has worked better for him than for Kamala.  Bourne out in the regional, cross-sectional data. 

The best advice is to do what I recommended that Kamala do post-DNC.  Move minimally; drain the clock.  This goes for JD as well, who by most accounts performed well in his debate.  There is a straight risk of misadventure if not.  Miscommunication, scandal, some sort of health embarrassment.  Things can happen to anyone.  Plus of course who needs another close attempt on life?  To some degree just playing on a golf course is better than another engineered debate, or ambushed interview.  And while I don’t advise it anymore, his relentless efforts in battleground Pennsylvania in early October have clearly paid off but could now be hitting an exhaustive ceiling.  Things will be very close, prepare for a hung result and after-hours ballot counting, for many nights after.

salil statistics [10k+ books sold, 36m reads, 1/4m follows]

buy copula narrative 

search products w.i. a year: audios, math w mira👶, and coding&logic

follow via RSS or e-mail or Amazon

free subscription to Salil Statistics