Rohan: A poll that’s too good to be true?

I wanted to write this after Steve Kates lamented why the Essential poll showed 61% support of Dan Andrews, in his The Melbourne Syndrome Revisited post. I said at the time something along the lines of, the polls are manipulated in such a way so they’re used as a weapon to whack the neigh-sayers. They’re in effect a political cancel culture attack vehicle and outright propaganda. I still hold that view.

However, there’s real world data in the public domain, that since its inception in 2004, is batting at 100%. I was enlightened just after the 2016 US Presidential election, by Jim Stewart. Jim is a Search Engine Optimisation expert. His company consults for businesses, getting their websites to rank in Google search engine results either organically or using Adwords. Just after Trumps win, he posted a video looking at the Google Trends results. In that video, he stated that Google Trends has correctly picked the winner in every Presidential election since 2004, and indeed the election in every Western country since. It correctly picked Scott Morrison’s win in last year’s federal election. Even the two elections contested and won by the Canadian champion of blackface, Justin Trudeau. To illustrate the point, I suggest you watch this 9:12 minute video where Jim breaks down the data, analysing the 2019 Federal Election.

Google collects the input of search phrases every time you Google something. It’s a search phrase popularity contest. This information can be used to write content for a website, especially in an e-commerce setting. But if you have smart election campaign staff, it can also be used to gauge how well your campaign is going and what issues you need to address.

Now that we’re in the last 2 weeks of the Presidential election, let’s compare the last 2 weeks of the 2016 Presidential election campaign to the current previous 2 weeks. I’ve used the search terms Hillary Clinton (blue team) and Donald Trump (red team). Then done the same for this campaign: Joe Biden (Blue Team) and Donald Trump (red team). Note the filter used to select results exclusively from where it counts, in the United States.

When you look at the averaged interest for the period on the left, that is very close to the actual margin of electoral colleges won by Trump in 2016. It really is very good indicator.

Now look at the last 2 weeks to date of the Biden vs Trump election race in the current race.

Note the average interest for the period. At this point Google Trends shows Trump is #winning comfortably and by a much healthier margin than 2016. If you then look at every other issue facing this election, levels of interest, policy positions, law and order, Hunter Biden etc, you know the democrats are looking more and more like charcoaled toast.

#trumpslidepending

This entry was posted in Elections, Guest Post. Bookmark the permalink.

28 Responses to Rohan: A poll that’s too good to be true?

  1. Sinclair Davidson

    Closer to election day I will open up a thread for explicit election predictions. Not that either Biden or Trump will win, but also the number of electoral votes they’ll pick up. So please keep your powder dry on that point.

  2. RobK

    Google could save us the cost of elections 😀

  3. H B Bear

    After the World Cup I only use octopuses.

  4. Aethelred

    So what is Dear Leader’s approval?

    How many punters have been googling “escape from Melbourne”

  5. Old School Conservative

    Great article and also a clever reference to Jacinda Ardern.

  6. cuckoo

    The black swan remains that Twitter poll the Age – the Age – ran on Andrews’ ‘roadmap’. By the time it closed it had a not shabby response of over 112k, with ‘disapprove’ running at 81 percent.

  7. Cynic of A

    OK, but can Google really be trusted to provide this data accurately and without bias?
    Is Google capable of the dishonesty to alter these results to suit their agenda, which is Biden?
    Recent events would seem to indicate, “No, and Yes.”

  8. Professor Fred Lenin

    My take on polls is always ,feed in crap and you get gold out ,yeah right comrades .
    Biden leading ! and he amd harris cam]n barely get 50 people at their meetings . the security at Trumps rallies outnumbers the biden harris crowds by a big margin .

  9. Judge Dredd

    It will be a Trumpslide even bigger than 2016. You’d almost have to be watching the MSM 24×7 to be indoctrinated to think Biden will win at the present time.
    The US economy, even with Covid, has rebounded hugely which alone guarantees a Trump win. Trumps rallies are huge versus Biden’s impotent little group chats. And lastly Hunter’s laptop full of the usual wicked stuff. … No one wants Biden to win unless you’ve been deceived or you’re the deceiver.

  10. calli

    Fascinating analysis. I presume the spike was Trump’s Covid diagnosis?

  11. Bad Samaritan

    On Sunday the 18th Joe was $1.50 and Donald $3 on Betfair. At that point there was about $85 million for Joe and about $100 million for Donald, which I pointed out made no sense, except that it indicated the Joe fans were being skinned by the Donald fans. In simple terms the Joe-supporters were offering the $3 and should have been winding the price in as the money came…but they weren’t.

    OK, today it’s still $1.50 and $3. There is now about $137 million on Joe and $153 million on Donald. In other words, each has had about the same amount ($52-$53 million) bet on each since then but the price has not changed.. The Joe fans are still offering Donald at $3 despite that much money coming for him, and still accepting $1.50 when it is offered.

    Currently a Donald win will cost the Joe fans about $225 million., while a Joe win would cost the Donald fans about $60 million. This means the Joe fans are really in a deep hole. The outsider winning (Donald in this case) should never be the big liability result. Donald fans are way too smart for the Joe dopes and this proves it. Trump to win easily.

  12. Roger

    I said at the time something along the lines of, the polls are manipulated in such a way so they’re used as a weapon to whack the neigh-sayers.

    I don’t think polls generally are manipulated (not electoral polls, anyway; polls on particular issues may be another matter).

    It’s just that telephone polling no longer provides accurate data on voter intentions across the spectrum of any given electorate for several reasons.

  13. min

    I heard an Australian professor who analyses data from social media to predict election results who has 100% accuracy including Scomo, Brexit , 2016 Trump .Trudeau etc
    Polls are rigged through sampling and methodology . If 1000 are asked in a place with 25 million with diverse demographic and ethnic populations what results would you expect.?

  14. Roger

    Seven days out from our state election Google Trends has Palaszczuk heading for a landslide win.

    Just last week she was just ahead of Frecklington, which seemed to be what the polling was saying., although up until the opening of hte campaign the LNP had been in front since March.

    The problem with their data is it’s only taken from Brisbane and the GC, which obviously doesn’t reflect the QLD electorate.

  15. Lee

    It will be a Trumpslide even bigger than 2016.

    I certainly hope so.
    I don’t think a Democrat win is in Australia’s best interest at all. Nor America’s.

  16. Roger

    If 1000 are asked in a place with 25 million with diverse demographic and ethnic populations what results would you expect.?

    Min, is there any evidence such selective polling is happening? If it was and is came out it would seriously damage the reputation of the polling firm and probably put them out of business.

    What pollsters have admitted is that response rates are down considerably from where they were two decades ago. People just don’t pick up the phone or don’t even have a landline anymore. And if they do they may not tell the truth to the pollster. This means meaningful data takes longer to obtain and is more expensive, perhaps leading to short cuts being taken.

  17. Bad Samaritan

    Further to the above post at 5.12pm. That should’ve begun “On Sunday the 11th

  18. FelixKruell

    When the difference between winning and losing is less than 100,000 votes in a few states, of course polls are dodgy. Google (and octopi) not much better.

  19. Another Ian

    “Professor Fred Lenin
    #3630120, posted on October 23, 2020 at 4:37 pm

    My take on polls is always ,feed in crap and you get gold out ,yeah right comrades .
    Biden leading ! and he amd harris cam]n barely get 50 people at their meetings . the security at Trumps rallies outnumbers the biden harris crowds by a big margin .”

    I saw a comment somewhere that the line at any one Porta-Potty at Trump meetings exceed “bidarris” total attendance

  20. ACTOldFart

    The plural of octopus is, according to SOED, octopuses, or octopodes

  21. nb

    ‘neigh-sayers’
    Polls are just a whole lot of horsing around.

  22. Another Ian

    If a poll like this actually works I guess we can trust Google to fix it?

  23. nb

    RobK, #3630072: ‘Google could save us the cost of elections 😀’
    They’re trying to. Along with the MSM, Twitter, Facebook et al, and especially our friends the Chinese. 😉

  24. Chris M

    The US left is going the South American way with hyperactive mobs, corrupt dictatorial leaders and coups.

    At least Trump could oblige them by having the failed coup plotters machine-gunned.

    BTW Trump = 37 states, plus or minus 1.

  25. Roger

    1 day later Google now has Frecklington ahead of Palaszczuk.

    It’s nothing if not volatile.

Comments are closed.