Polling analyses: a few thoughts

So, we are at the last weekend before Election Day. Coming out are all of these last-minute polls and polling analyses and the concomitant verdict of who is going to win. The ones making the final forecast seems to always preface their declarations, “I don’t believe in the polls…but…”

Why don’t they believe in the polls? Probably because of what transpired on 2016 election when all polls pointed to Hilary Clinton winning and election night gave a totally different shocking result. But there is nothing else on hand. We only have polls to give us a peek at what the voters may be thinking.

Imperfect as those polls are, we still cling to them for answers.

I’m seeing a lot of people “confidently” declaring Trump will win or Kamala will win, but the polls that I see seem to show one of two points variances, all within the margin of error.

Then there are the analyses…

Analysis 1: Red Wave polls

Within the last week or two weeks, overall polls seemed to indicate that Trump was starting to trend upward, meaning he was gaining ground and overcoming Kamala in the national polls. Lots of anxiety ensued.

However, somebody said that a series of polls from conservative pollsters, and not very high-quality ones either, were dumped in the last two weeks or so, polluting the overall national level with poor polling results. The red wave polls were so numerous that they started to drive the trends in Trump’s favor.

That’s one way of manipulating the national narrative, even if the end polling result on election day does not pane out as projected.

Why would anybody do this? What’s the point if you are still going to lose? I think it’s something called psychological warfare. They do this in the hopes of getting the opposing team to feel defeated and not turn out the vote.

That’s an interesting way of playing the game…and maybe not honest.

So, one type of analysis of the polls is to see if a whole bunch of polls were just dumped into the national arena by a particular group of pollsters, such as the low-quality pollsters trying to drive a certain conversation as part of a psychological warfare.

Analysis 2: Mainline versus internet reach

One analyst postulated that respondents were still mainly from the mainline rather than cellphones or social media. If that is the case, then the polling analyses are missing a whole set of potential voters who could swing the election in one way.

If that is true (respondent mainly those using mainline), then we don’t know dipsquat. We are really operating in the dark. People using mainline phones tend to be older; cellphones tend to be dominated by the youngsters.

So, does that mean that Kamala might have the edge here because the younger generations are missing from the polls? Not necessarily. It could be that there is a hidden block of Trump supporters in those younger generations. Have you looked at those rallies? They look kind of young.

I wouldn’t listen to those who are confident that one or the other candidate will win because we simply don’t know. That one or two margin difference may or may not mean a close election if we are missing a whole class of voters.

Analysis 3: Surface polling analyses of early polling numbers are meaningless

According to somebody’s analysis, and it makes a lot of sense, just looking at the early polling numbers without looking at state rules on early voting is pointless. Unless you know state’s policies on early voting and mail in voting, looking at the early voting numbers (the participation rate versus prior years) alone does not mean anything.

That same person was using Georgia and Pennsylvania as examples to explain her position. She said it was difficult to derive any meaning in the early voting or mail in results. Georgia’s early voting just exploded this year (and I still don’t know what that means) and Pennsylvania apparently was anemic. I guess somebody tried to attribute some meaning to the supposedly anemic showing in Pennsylvania relative to Georgia. However, you can’t ascribe any meaning or any analysis because Pennsylvania does not have early voting and Georgia does. You are comparing apples to oranges, so to speak.

I have no clue as to what that means for who is winning.

In other words, she said to dig into the numbers to understand what they mean. We won’t know Pennsylvania’s results until Election Day (and maybe later) because the state only has either mail in or in person on Election Day. Furthermore, counting doesn’t start until Election Day so the likelihood of knowing the results on Election Day is pretty slim, unless either candidate has a blowout (just receive so much votes) that the result is uncontestable.

Unfortunately, I think the vote will be a LOT closer in Pennsylvania.

But you need to dig into the numbers to understand them. Polling analyses need to go deeper.

Analysis 4: Don’t believe the PolyMarket hype (addendum: added 11/3/2024)

There’s a new type of betting site where one bet cryptocurrencies on any number of betting events: weather events, sporting events, Grammy awards, elections. I learned about the crypto one a few days ago and intended to include in the post last night.

In the theory of the crowds predicting better than a single person, one would think that betting sites, similar to stock markets, would indicate results much better than any single prognosticator. PolyMarket, the crypto site, has been indicating that Trump would win.

However, a reporter cautioned relying on the conclusions of that betting market and gave good reasons for the caution:

• US investors cannot go on the site due to the owner of the betting site setting it up offshore to remain out of reach of US regulators. Apparently, there is some kind of rule about not having gambling on games or election results. Consequently, the bettors are foreigners.
• The US has some foreign enemies so trying to drive the results of the site to some certain outcome, such as implying Trump will win, might appeal to such enemies (say Russia).
• It apparently relatively easy for the wealthy bettor to drive the direction of the results; the reporter said it is the price of a used car.
• Finally, we don’t know who the foreign investors populating this crypto site.

Needless to say, this crypto site may be unreliable for analyses.

Analysis 5: The mirages (addendum, 11/3/2024)

Watching the polls or watching the election returns could lead you astray due to mirages. They are often called red mirages or blue mirages and stems from the timing of receipt of the ballots or the counts.

This NBC site does an excellent job of describing the instances of these mirages and how to interpret the returns on Tuesday night. I’m not going to go through the particular details because this site does a very good job of explaining the phenomenon.

Analysis 6: The Iowa Selzer poll (addendum, 11/3/2024)

Videos discussing this Iowa poll came out last night after I had done this post. I decided I would add in this analysis for posterity.

When I got off of my machine and went looking for news, a bunch of videos scrolled about this news breaking poll out of Iowa and I was left that this was a shocking phenomenon, just from the video titles. I went first to the normal analysts’ video to get their take and the hosts were practically bouncing out of their chairs and their smiles were huge.

Okay, they got news that were making them ecstatic. I was like “Calm down. Don’t get your hopes up.”

This pollster is supposed to be the gold standard for Iowa because for at least the last decade, her last poll just before the election day has mostly been on the dot. She missed once. So, her poll predicted that Trump would win Iowa when all other polls said Clinton. She turned out to be correct.

And she has done similar feat.

However, she is polling Iowa, and the shocking results may be only applicable to Iowa. Everyone was trying to extend Iowa’s results to other states and that may not be right. One guy was basically turning most red states blue. It was ridiculous and it was that ridiculousness that makes me roll my eyes.

We have to be very careful of extending Iowa results to other states, even if it is tempting to consider that something is going on with the voters to make them flip from red to blue.

Analysis 7: NONE of these polls take into consideration the possibility of Trump politicians gaming the system (addendum, 11/3/2024)

The polls do not capture the plans that Trump politicians have up their sleeves. Trump said that he has a “secret” with Mike Johnson which could impact the election results. Polls will never capture this.

Polling analyses: the end is near

At the beginning of this year, I dreaded the year because I knew we would be facing this election. I feared the potential violence and this past week I saw at least 3 videos discussing 3 separate instances of quasi-violence at the polls. No polling analyses will contemplate real world actions such as violence having an impact on behavior and choices. Maybe these instances of quasi-violence will move people to stay home and thus the election swings one way. Polling does not capture that.

I think polling also may not capture the impact of candidate behavior on voter preferences, especially during the last week.

I’m going to be oblique here, but I am referencing a particular candidate talking about some great golfer’s body part and then later, making some allusions to some sexual act with a mic. With this particular candidate, there seems to be a lot of obsessions with sex and I can’t help but question if he’s actually a closet gay. Not that that should matter but I do wonder if his possible cognitive decline means he cannot control himself and is projecting his real tendencies (gay tendencies). Polling analyses will not capture these incidences and their impact on voter choice. The reason for voters’ choice may be these types of “unsaid” thinking.

Yes, the end is near, but the tension will go on for a couple of more months through legal maneuvers. This is the part I’m dreading: the high stakes tension while the legal minds battle. This could be the constitutional crisis. Yeah, what I feared at the beginning of the year is here.