Users Online Now: 1,553 (Who's On?) Visitors Today: 400,609 Pageviews Today: 702,769 Threads Today: 165 Posts Today: 2,986 09:45 AM

Absolute BS Crap Reasonable Nice Amazing

# DEBUNK THIS: Google trends of various elections: Is google more accurate than polls? (It has predicted the winner in all the last 4 elections)

ToSeek

User ID: 9653749
United States
08/24/2020 12:06 PM

Report Abusive Post
Re: DEBUNK THIS: Google trends of various elections: Is google more accurate than polls? (It has predicted the winner in all the last 4 elections)
I've found an exception! Ted Cruz beat Beto O'Rourke in the 2018 Texas Senate election even though O'Rourke was well ahead according to Google: [link to trends.google.com (secure)]

User ID: 78331155
08/24/2020 12:10 PM
Report Abusive Post
Re: DEBUNK THIS: Google trends of various elections: Is google more accurate than polls? (It has predicted the winner in all the last 4 elections)
...

Nope!
filter by Kansas (click on the state in the map)
And set the proper date .. it was in April 2017!

Totally legit!

Didn't know you could break it down by state. I used the January date because you are going 3 months out ahead of the actual election for your analysis, so that is what I was doing.
Quoting: TrustNoOneKS

The election happened in 2017

It makes sense to filter by the state to avoid the noises of the rest of the country.

The 2018 date was for the 2018 election. I used 2 trends for 2 different elections. I used 2017 for the 2017 special election.
Quoting: TrustNoOneKS

Filtered by Kansas & end date 15th of November 2018

very close -

Filtered by Wichita

still very close according to google data

The brownback/Paul David case is over 830,000 votes in total

Obviously the lower the population the less precise google is.

It's a good exercise to test the limits of the system!
I'm a J & proud zio.

FE Challenge: Provide a formula which calculates the exact distance between 2 GPS coordinates that does not use the Earth's radius of 6,371 km in it's assumptions

Anonymous Coward
User ID: 78246572
United States
08/24/2020 12:12 PM
Report Abusive Post
Re: DEBUNK THIS: Google trends of various elections: Is google more accurate than polls? (It has predicted the winner in all the last 4 elections)
If you cherry pick the search terms you can get the outcome you want.
Quoting: Anonymous Coward 78715086

Didn’t like what you saw huh? Had to get nasty? Lol November going to be rough month for you
Quoting: lionright

There will be a whole lot of he/shes with brain hemorrhoids once Trump wins again.
Anonymous Coward
User ID: 54562618
United States
08/24/2020 12:12 PM
Report Abusive Post
Re: DEBUNK THIS: Google trends of various elections: Is google more accurate than polls? (It has predicted the winner in all the last 4 elections)
(Update: It seems you can use this to detect electoral fraud)

2004 election

2008 election:

2012 election:

2016 election:

2020 election: (past 90 days)

This is going to cause major butthurt

User ID: 78331155
08/24/2020 12:25 PM
Report Abusive Post
Re: DEBUNK THIS: Google trends of various elections: Is google more accurate than polls? (It has predicted the winner in all the last 4 elections)
I've found an exception! Ted Cruz beat Beto O'Rourke in the 2018 Texas Senate election even though O'Rourke was well ahead according to Google: [link to trends.google.com (secure)]
Quoting: ToSeek

Beto vs Cruz filtered by Texas shows 50/50

In truth it was 50.9 vs 48.3

I don't think we can expect Google to be more precise than 3%

Last Edited by JADR+ on 08/24/2020 12:30 PM
I'm a J & proud zio.

FE Challenge: Provide a formula which calculates the exact distance between 2 GPS coordinates that does not use the Earth's radius of 6,371 km in it's assumptions

TrustNoOneKS

User ID: 76946759
United States
08/24/2020 12:53 PM
Report Abusive Post
Re: DEBUNK THIS: Google trends of various elections: Is google more accurate than polls? (It has predicted the winner in all the last 4 elections)
Well, I hate to ruin your trend again with another race, but Ron Estes (R) won over James Thompson (D) for the House of Representatives in 2018 in Kansas. Once more, the loser, had far more than the winner on the trends:

Interestingly, he beat the same guy in the special elections in April 2017 to replace Pompeo, and the trend also favored Thompson then as well:

Quoting: TrustNoOneKS

Bownback still loses if you filter by Kansas.

(I think it's a reasonable assumption to filter the data by the state in which people vote)

Okay, so I did a new search for Brownback and Davis running it through the day after the election.

I think I might be seeing why Davis would be trending. When you go all the way down to the bottom under Paul Davis, you see a result that says "Paul Davis strip club" under related searches. Is this what is making the numbers wonky on this one election because of his negative press being such a big deal over the strip club debacle?

I know you want to say cheating, but I really don't think there is clear evidence of that and it's not quite right to throw that out there when there has been no actual evidence of it.

The wonky result makes more sense when you throw in a "sex" angle to things making it getting more searches done on it. Out of curiosity to test this theory, I did a Paul Davis strip club search to see what that is all about. I see that the articles are dating to September 20th when you see a big uptick on that day in particular for Paul Davis. Interestingly, the strip club incident is still being brought up in 2018. Another big uptick day for Paul Davis is September 23rd and on that day there was controversy over a Supreme Court justice hosting a fund raising dinner for him.

So, could the actual results not be because people are actually interested in voting for the guy but because people were looking into his shady dealings?
I Want To Believe

User ID: 78331155
08/24/2020 01:14 PM
Report Abusive Post
Re: DEBUNK THIS: Google trends of various elections: Is google more accurate than polls? (It has predicted the winner in all the last 4 elections)
Well, I hate to ruin your trend again with another race, but Ron Estes (R) won over James Thompson (D) for the House of Representatives in 2018 in Kansas. Once more, the loser, had far more than the winner on the trends:

Interestingly, he beat the same guy in the special elections in April 2017 to replace Pompeo, and the trend also favored Thompson then as well:

Quoting: TrustNoOneKS

Bownback still loses if you filter by Kansas.

(I think it's a reasonable assumption to filter the data by the state in which people vote)

Okay, so I did a new search for Brownback and Davis running it through the day after the election.

I think I might be seeing why Davis would be trending. When you go all the way down to the bottom under Paul Davis, you see a result that says "Paul Davis strip club" under related searches. Is this what is making the numbers wonky on this one election because of his negative press being such a big deal over the strip club debacle?

I know you want to say cheating, but I really don't think there is clear evidence of that and it's not quite right to throw that out there when there has been no actual evidence of it.

The wonky result makes more sense when you throw in a "sex" angle to things making it getting more searches done on it. Out of curiosity to test this theory, I did a Paul Davis strip club search to see what that is all about. I see that the articles are dating to September 20th when you see a big uptick on that day in particular for Paul Davis. Interestingly, the strip club incident is still being brought up in 2018. Another big uptick day for Paul Davis is September 23rd and on that day there was controversy over a Supreme Court justice hosting a fund raising dinner for him.

So, could the actual results not be because people are actually interested in voting for the guy but because people were looking into his shady dealings?
Quoting: TrustNoOneKS

The lower the resolution ... the less precises it is, and the more 'the noise' has some effect

If I filter it down to Wichita & add 7 extra days after the election - I get (12/10) 55/45 for Brownback

The real results were 49/46 for BrownBack.

Notice how basic the pattern is.

I take back what I said about his voter fraud

I think google reaches it's limits at those local levels.

(which is to be expected)

However you can bet your bottom dollars that on a national level, with 100 million voters or more, it will be right on the money.

Last Edited by JADR+ on 08/24/2020 01:15 PM
I'm a J & proud zio.

FE Challenge: Provide a formula which calculates the exact distance between 2 GPS coordinates that does not use the Earth's radius of 6,371 km in it's assumptions

ToSeek

User ID: 9653749
United States
08/24/2020 01:23 PM

Report Abusive Post
Re: DEBUNK THIS: Google trends of various elections: Is google more accurate than polls? (It has predicted the winner in all the last 4 elections)
I've found an exception! Ted Cruz beat Beto O'Rourke in the 2018 Texas Senate election even though O'Rourke was well ahead according to Google: [link to trends.google.com (secure)]
Quoting: ToSeek

Beto vs Cruz filtered by Texas shows 50/50

In truth it was 50.9 vs 48.3

I don't think we can expect Google to be more precise than 3%

You're using "Beto" and "Cruz." How many Cruzes are there in Texas? If you use their full names, it's a big lead for O'Rourke, even in Texas: [link to trends.google.com (secure)]

User ID: 78331155
08/24/2020 01:32 PM
Report Abusive Post
Re: DEBUNK THIS: Google trends of various elections: Is google more accurate than polls? (It has predicted the winner in all the last 4 elections)
I've found an exception! Ted Cruz beat Beto O'Rourke in the 2018 Texas Senate election even though O'Rourke was well ahead according to Google: [link to trends.google.com (secure)]
Quoting: ToSeek

Beto vs Cruz filtered by Texas shows 50/50

In truth it was 50.9 vs 48.3

I don't think we can expect Google to be more precise than 3%

You're using "Beto" and "Cruz." How many Cruzes are there in Texas? If you use their full names, it's a big lead for O'Rourke, even in Texas: [link to trends.google.com (secure)]
Quoting: ToSeek

Scroll down to related queries:

cruz beto 100
beto 100
ted cruz 100
beto vs cruz 47
ted cruz beto 28
cruz azul 25
polls 24
cruz debate 23
cruz polls 23
beto cruz debate 20
polls cruz beto 17
santa cruz 12
beto vs cruz polls 11
beto vs ted cruz 11
cruz poll 10
election 9
cruz election 9
beto and cruz 9
beto cruz poll 9
cruz azul vs 7
ted cruz polls 6
ted cruz debate 6
texas senate 5
beto cruz results 5
texas beto vs cruz 5

RISING
beto cruz results Breakout
beto vs cruz debate Breakout
cruz or beto Breakout
beto vs cruz results Breakout
beto vs cruz poll Breakout
texas senate race Breakout
beto vs cruz policies Breakout
beto cruz debate schedule Breakout
ted cruz vs beto policies Breakout
ted cruz vs beto debate Breakout
texas election results Breakout
election results 2018 Breakout
midterm elections Breakout
who won beto cruz debate Breakout
who won cruz or beto Breakout
trump cruz rally Breakout
who won beto vs cruz Breakout
beto o rourke vs ted cruz Breakout
ted cruz yard signs Breakout
kimmel vs cruz Breakout
whos winning beto or cruz Breakout
cruz beto poll numbers Breakout
who is winning beto or cruz Breakout
beto vs cruz stances Breakout
election day 2018 Breakout

Last Edited by JADR+ on 08/24/2020 01:38 PM
I'm a J & proud zio.

FE Challenge: Provide a formula which calculates the exact distance between 2 GPS coordinates that does not use the Earth's radius of 6,371 km in it's assumptions

TrustNoOneKS

User ID: 76946759
United States
08/24/2020 01:48 PM
Report Abusive Post
Re: DEBUNK THIS: Google trends of various elections: Is google more accurate than polls? (It has predicted the winner in all the last 4 elections)
So, I retested your theory on all the trends in the OP. You just use last names. I used full names. I also adjusted to stop the day after the election because whatever trends the day after the election shouldn't affect the voting and started with the June 1st date that you were using.

I found a wildly different comparison with using John Kerry versus George Bush. And yes, we should probably use first names because people could actually search for a bush, as in a shrub.

I also found a discrepancy with Barack Obama versus John McCain. McCain leads:

2012 holds true (Obama/Romney) using full names

2016 holds true (Clinton/Trump) using full names

2020 holds true (Trump/Biden) using full names
I Want To Believe

User ID: 78331155
08/24/2020 01:53 PM
Report Abusive Post
Re: DEBUNK THIS: Google trends of various elections: Is google more accurate than polls? (It has predicted the winner in all the last 4 elections)
So, I retested your theory on all the trends in the OP. You just use last names. I used full names. I also adjusted to stop the day after the election because whatever trends the day after the election shouldn't affect the voting and started with the June 1st date that you were using.

I found a wildly different comparison with using John Kerry versus George Bush. And yes, we should probably use first names because people could actually search for a bush, as in a shrub.

I also found a discrepancy with Barack Obama versus John McCain. McCain leads:

2012 holds true (Obama/Romney) using full names

2016 holds true (Clinton/Trump) using full names

2020 holds true (Trump/Biden) using full names
Quoting: TrustNoOneKS

If you put John Kerry & George Bush- you eliminate all the people who search for "Bush vs Kerry"
People like to use short hand when they search

Bush vs Kerry search terms:

john kerry 99
bush kerry 74
bush 73
george bush 14
kerry and bush 14
kerry edwards 9
kerry poll 9
kerry president 9
debate 9
kerry 2004 8
heinz kerry 8
polls 7
kerry campaign 7
kerry polls 7
kerry bush debate 7
teresa kerry 7
bush kerry polls 5
vote kerry 5
teresa heinz kerry 5
bush kerry this land 5
this land 4
kerry cartoon 4
kerry vietnam 4
kerry for president 4

RISING
bush kerry this land Breakout
this land Breakout
bush kerry this land is your land Breakout
kerry concession Breakout
bush kerry song Breakout
kerry concession speech Breakout
bush and kerry cartoon Breakout
bush kerry parody Breakout
kerry cambodia Breakout
bush kerry jib jab Breakout
bush kerry animation Breakout
bush kerry flash Breakout
debate transcript Breakout
democratic national convention Breakout
bush kerry this land is my land Breakout
bush kerry debate transcript Breakout
kerry concedes Breakout
gallup poll Breakout
jibjab bush kerry Breakout
kerry spot Breakout
exit polls Breakout
bush kerry debate polls Breakout
kerry tan Breakout
kerry buddy icons Breakout
kerry acceptance speech Breakout

George Bush vs John Kerry (search terms)

bush 100
george bush 71
john kerry president 24
john edwards 19
george w. bush 19
kerry edwards 19
john kerry john edwards 18
john kerry campaign 17
john kerry 2004 15
john kerry biography 15
john kerry pictures 12
george bush and john kerry 12
election 11
george w bush 11
john kerry for president 11
john kerry vietnam 11
john kerry senate 11
kerry for president 10
john kerry speech 9
senator john kerry 9
vote john kerry 9
john kerry voting record 8
president bush 8
john kerry website 8
john kerry picture 7

RISING
john kerry football Breakout
john kerry t-shirts Breakout
john kerry humor Breakout
stem cell research Breakout
john kerry nasa Breakout
john kerry purple hearts Breakout
george bush john kerry cartoon Breakout
laura bush Breakout
john kerry discharge Breakout
john kerry in vietnam Breakout
where was john kerry born Breakout
john kerry war crimes Breakout
the ultimate john kerry ad Breakout
john kerry honorable discharge Breakout
johnkerry.com Breakout
john kerry concession speech Breakout
swift boat 1150%
john kerry facts 600%
george bush biography 600%
george bush and john kerry 550%
presidential debate 550%
dick cheney 550%
john kerry cartoons 500%
picture of john kerry 500%

Last Edited by JADR+ on 08/24/2020 01:56 PM
I'm a J & proud zio.

FE Challenge: Provide a formula which calculates the exact distance between 2 GPS coordinates that does not use the Earth's radius of 6,371 km in it's assumptions

User ID: 78331155
08/24/2020 02:05 PM
Report Abusive Post
Re: DEBUNK THIS: Google trends of various elections: Is google more accurate than polls? (It has predicted the winner in all the last 4 elections)
So, I retested your theory on all the trends in the OP. You just use last names. I used full names. I also adjusted to stop the day after the election because whatever trends the day after the election shouldn't affect the voting and started with the June 1st date that you were using.

I found a wildly different comparison with using John Kerry versus George Bush. And yes, we should probably use first names because people could actually search for a bush, as in a shrub.

I also found a discrepancy with Barack Obama versus John McCain. McCain leads:

2012 holds true (Obama/Romney) using full names

2016 holds true (Clinton/Trump) using full names

2020 holds true (Trump/Biden) using full names
Quoting: TrustNoOneKS

The context is given by the place and time & frequency of the searches...

Bush/ Kerry in the US in an election year - when Bush & Kerry are running for president makes it's highly unlikely for your search to be about a bush named Kerry.

Google is smart .. no point in being too specific with your search as you only narrow the overall picture.

Last Edited by JADR+ on 08/24/2020 02:10 PM
I'm a J & proud zio.

FE Challenge: Provide a formula which calculates the exact distance between 2 GPS coordinates that does not use the Earth's radius of 6,371 km in it's assumptions

SerenaSeesAll

User ID: 78926188
United States
08/24/2020 02:08 PM

Report Abusive Post
Re: DEBUNK THIS: Google trends of various elections: Is google more accurate than polls? (It has predicted the winner in all the last 4 elections)

This was for the last FIVE years
Trump 95%
Biden 5%
=============================================================​

Noticing Miracles Is My Hobby

The Bible - When you carry the Bible, Satan has a headache, when you open it, he collapses, when he sees you reading it, he loses his strength, AND when you stand on the Word of God, Satan can't hurt you!

User ID: 78331155
08/24/2020 02:12 PM
Report Abusive Post
Re: DEBUNK THIS: Google trends of various elections: Is google more accurate than polls? (It has predicted the winner in all the last 4 elections)

This was for the last FIVE years
Trump 95%
Biden 5%
Quoting: SerenaSeesAll

Not a reversible trend! (not in 3 months anyway)

Last Edited by JADR+ on 08/24/2020 02:12 PM
I'm a J & proud zio.

FE Challenge: Provide a formula which calculates the exact distance between 2 GPS coordinates that does not use the Earth's radius of 6,371 km in it's assumptions

TrustNoOneKS

User ID: 76946759
United States
08/24/2020 02:24 PM
Report Abusive Post
Re: DEBUNK THIS: Google trends of various elections: Is google more accurate than polls? (It has predicted the winner in all the last 4 elections)
I need to get to bed, but as I was trying to point out with the strip club debacle, a search does not necessarily mean a person is interested in voting for a person just because they search for that person. I've done lots of searches on people I never voted for just to find information to cite.

Also from their own site:

"A spike in a particular topic does not reflect that a topic is somehow 'popular' or 'winning,' only that for some unspecified reason, there appear to be many users performing a search about a topic."

As a further example, I took your searches for the OP and just put them for Kansas instead of the whole U.S.

2004 Kansas leaned towards Bush on the trends chart (electoral went for Bush)

2008 Kansas leaned toward Obama on the trends chart (electoral went for McCain)

2012 Kansas leaned toward Obama on the trends chart (electoral went for Romney)

2016 Kansas leaned toward Trump on the trends chart (electoral went for Trump)

2020 Kansas leans toward Trump (electoral will most likely go to Trump)

Just because Obama was top search here in 2008 and 2012 does not mean the majority ended up voting for him. The electoral still went to the other candidate even though your chart shows him winning on the trends, even in Kansas.

Last Edited by TrustNoOneKS on 08/24/2020 02:26 PM
I Want To Believe
Anonymous Coward
User ID: 79301306
08/24/2020 02:25 PM
Report Abusive Post
Re: DEBUNK THIS: Google trends of various elections: Is google more accurate than polls? (It has predicted the winner in all the last 4 elections)
(Update: It seems you can use this to detect electoral fraud)

Google Trends is based directly on work I did from 1993 to 2004, it's a copy of my website I had up. The Google
engineer that copied it is Bruno Bowden.

That being said, I met Bruno in 2008 at Google HQ and
he confirmed that Google "massages" their data for
certain reasons.

I use Google Trends but I myself have seen it drastically
manipulated twice. Values in January were the opposite
in Feb, etc.

.
Anonymous Coward
User ID: 78224572
United States
08/24/2020 02:31 PM
Report Abusive Post
Re: DEBUNK THIS: Google trends of various elections: Is google more accurate than polls? (It has predicted the winner in all the last 4 elections)
Not surprised i mean even hardcore leftists are not excited about Biden i mean his slogan might as well be Orange man bad instead of build back better...
Anonymous Coward
User ID: 79301306
08/24/2020 02:34 PM
Report Abusive Post
Re: DEBUNK THIS: Google trends of various elections: Is google more accurate than polls? (It has predicted the winner in all the last 4 elections)
(Update: It seems you can use this to detect electoral fraud)

Google Trends is based directly on work I did from 1993 to 2004, it's a copy of my website I had up. The Google
engineer that copied it is Bruno Bowden.

That being said, I met Bruno in 2008 at Google HQ and
he confirmed that Google "massages" their data for
certain reasons.

I use Google Trends but I myself have seen it drastically
manipulated twice. Values in January were the opposite
in Feb, etc.

.
Quoting: Anonymous Coward 79301306

run these two queries, "big data" and "hadoop" and compare
what you get to what I recorded 2014.

.
Anonymous Coward
User ID: 79041354
United States
08/24/2020 04:01 PM
Report Abusive Post
Re: DEBUNK THIS: Google trends of various elections: Is google more accurate than polls? (It has predicted the winner in all the last 4 elections)
Not surprised i mean even hardcore leftists are not excited about Biden i mean his slogan might as well be Orange man bad instead of build back better...
Quoting: Anonymous Coward 78224572

No shit really? 'Build back better is second grade grammar. Further verifies his loss of mental acuity.
Anonymous Coward
User ID: 78013725
08/24/2020 09:12 PM
Report Abusive Post
Re: DEBUNK THIS: Google trends of various elections: Is google more accurate than polls? (It has predicted the winner in all the last 4 elections)

User ID: 78331155
08/24/2020 09:39 PM
Report Abusive Post
Re: DEBUNK THIS: Google trends of various elections: Is google more accurate than polls? (It has predicted the winner in all the last 4 elections)
I need to get to bed, but as I was trying to point out with the strip club debacle, a search does not necessarily mean a person is interested in voting for a person just because they search for that person. I've done lots of searches on people I never voted for just to find information to cite.

Also from their own site:

"A spike in a particular topic does not reflect that a topic is somehow 'popular' or 'winning,' only that for some unspecified reason, there appear to be many users performing a search about a topic."

As a further example, I took your searches for the OP and just put them for Kansas instead of the whole U.S.

2004 Kansas leaned towards Bush on the trends chart (electoral went for Bush)

2008 Kansas leaned toward Obama on the trends chart (electoral went for McCain)

2012 Kansas leaned toward Obama on the trends chart (electoral went for Romney)

2016 Kansas leaned toward Trump on the trends chart (electoral went for Trump)

2020 Kansas leans toward Trump (electoral will most likely go to Trump)

Just because Obama was top search here in 2008 and 2012 does not mean the majority ended up voting for him. The electoral still went to the other candidate even though your chart shows him winning on the trends, even in Kansas.
Quoting: TrustNoOneKS

Overall Obama won the popular vote in both elections.

You are nit picking I think ...

Yeah the accuracy breaks down, the smaller the data sample is.

Last Edited by JADR+ on 08/24/2020 09:40 PM
I'm a J & proud zio.

FE Challenge: Provide a formula which calculates the exact distance between 2 GPS coordinates that does not use the Earth's radius of 6,371 km in it's assumptions

Anonymous Coward
User ID: 79302971
08/24/2020 10:50 PM
Report Abusive Post
Re: DEBUNK THIS: Google trends of various elections: Is google more accurate than polls? (It has predicted the winner in all the last 4 elections)
TrustNoOneKS

User ID: 76946759
United States
08/25/2020 12:02 AM
Report Abusive Post
Re: DEBUNK THIS: Google trends of various elections: Is google more accurate than polls? (It has predicted the winner in all the last 4 elections)
So, I retested your theory on all the trends in the OP. You just use last names. I used full names. I also adjusted to stop the day after the election because whatever trends the day after the election shouldn't affect the voting and started with the June 1st date that you were using.

I found a wildly different comparison with using John Kerry versus George Bush. And yes, we should probably use first names because people could actually search for a bush, as in a shrub.

I also found a discrepancy with Barack Obama versus John McCain. McCain leads:

2012 holds true (Obama/Romney) using full names

2016 holds true (Clinton/Trump) using full names

2020 holds true (Trump/Biden) using full names
Quoting: TrustNoOneKS

The context is given by the place and time & frequency of the searches...

Bush/ Kerry in the US in an election year - when Bush & Kerry are running for president makes it's highly unlikely for your search to be about a bush named Kerry.

Google is smart .. no point in being too specific with your search as you only narrow the overall picture.

I wasn't implying a bush named Kerry and I think you know that. My point is that you change the goal post when things are brought to light that differ with what you want to find.

You say don't use first names and you clearly used first names on some and not on others. In fact, you used John McCain on the one which I found a discrepancy by putting in Barack Obama instead of Obama like you did. So, it's not relevant to use first names when you search Obama and John McCain, but when I search Barack Obama and John McCain, all of sudden first names don't matter?

One other point that I thought of when going to be bed, is that another reason for the importance of first names is how does the internet know when someone searches say Obama, you aren't ruling out people who search Michelle Obama and when people are searching Bush, without putting in a first name, you don't know if they are searching George W. Bush, George H.W. Bush, Jeb Bush, or Laura Bush.

Last Edited by TrustNoOneKS on 08/25/2020 12:05 AM
I Want To Believe
Anonymous Coward
User ID: 68762245
Australia
08/25/2020 12:04 AM
Report Abusive Post
Re: DEBUNK THIS: Google trends of various elections: Is google more accurate than polls? (It has predicted the winner in all the last 4 elections)

User ID: 78331155
08/25/2020 12:26 AM
Report Abusive Post
Re: DEBUNK THIS: Google trends of various elections: Is google more accurate than polls? (It has predicted the winner in all the last 4 elections)
So, I retested your theory on all the trends in the OP. You just use last names. I used full names. I also adjusted to stop the day after the election because whatever trends the day after the election shouldn't affect the voting and started with the June 1st date that you were using.

I found a wildly different comparison with using John Kerry versus George Bush. And yes, we should probably use first names because people could actually search for a bush, as in a shrub.

I also found a discrepancy with Barack Obama versus John McCain. McCain leads:

2012 holds true (Obama/Romney) using full names

2016 holds true (Clinton/Trump) using full names

2020 holds true (Trump/Biden) using full names
Quoting: TrustNoOneKS

The context is given by the place and time & frequency of the searches...

Bush/ Kerry in the US in an election year - when Bush & Kerry are running for president makes it's highly unlikely for your search to be about a bush named Kerry.

Google is smart .. no point in being too specific with your search as you only narrow the overall picture.

I wasn't implying a bush named Kerry and I think you know that. My point is that you change the goal post when things are brought to light that differ with what you want to find.

You say don't use first names and you clearly used first names on some and not on others. In fact, you used John McCain on the one which I found a discrepancy by putting in Barack Obama instead of Obama like you did. So, it's not relevant to use first names when you search Obama and John McCain, but when I search Barack Obama and John McCain, all of sudden first names don't matter?

One other point that I thought of when going to be bed, is that another reason for the importance of first names is how does the internet know when someone searches say Obama, you aren't ruling out people who search Michelle Obama and when people are searching Bush, without putting in a first name, you don't know if they are searching George W. Bush, George H.W. Bush, Jeb Bush, or Laura Bush.
Quoting: TrustNoOneKS

I was fine tuning as I went along:

The method that works is:

1) Last names - not full names - (or whatever is a simplest shortest way to identify the candidate . E.G: Beto is better than O'Rourke - Mccain is better than John or John Mccain)
2) Date range starts from beginning of campaign to 1 week after the election.
3) Always Filter by geographical area where election takes place
(Country, state, county etc..)

You do that, 9 out 10 times - the results match reality.

Note: the predictive system starts to breaks down when there are less than 500,000 voters, or when the election is too tight (within 3% of each other.)

- the time & place & who is running defines the context

in 2008 most searches where not about Michelle Obama

when people tipe Obama they expect Barrack, not Michelle.

in 2004 during election period, you can bet most searches about bush were related to the elections.

Using the full names reduces the data.

The more data there is (within context), the more it matches reality.

Last Edited by JADR+ on 08/25/2020 12:27 AM
I'm a J & proud zio.

FE Challenge: Provide a formula which calculates the exact distance between 2 GPS coordinates that does not use the Earth's radius of 6,371 km in it's assumptions

TrustNoOneKS

User ID: 76946759
United States
08/25/2020 01:42 AM
Report Abusive Post
Re: DEBUNK THIS: Google trends of various elections: Is google more accurate than polls? (It has predicted the winner in all the last 4 elections)
So out of further interest in this, I did some searches on articles on this. The funny part is this article:

How Google searches might be able to predict the 2020 Democratic race

"And that's where this amazing chart -- via Google's Simon Rogers -- comes in. It tracks Google search interest in 23 of the Democratic candidates for president from December 30, 2018, until June 16.

. . .

Now, searching for a candidate on Google isn't the same thing as voting for them. You could, for example, look up "Elizabeth Warren" because you don't want her to win and need some talking points to push back on your pro-Warren friends."

The funny thing is that the chart goes through January 2020 even though the article is dated June 2019, so the chart kept updating apparently even after the article first published.

If you go based on the final date of January 22, 2020, then we should be looking at Bernie Sanders as the current nominee, not Joe Biden. Joe Biden was ranking at 4th on the last date the chart took data.

So based on the above, I took your method of just using last names of Biden and Sanders and ran it first for the same dates as the article and Sanders wins:

I then also ran it until Sanders dropped out. Sanders still is winning:

Here's another article where someone used Google Trends to predict that Romney would beat Obama (something we all now know to not have happened):

Google Predicts Mitt Romney To Win Fall Election?

"My first step was to compare the two candidates for web search Interest within the United States over the last thirty days.

What I discovered is that, according to the Google service, Mitt Romney (in red) is clearly leading in the amount of raw search activity compared to Barack Obama.

. . .

The results show that Romney is currently leading Obama in every State, even States you can assume are "blue" or typically Democratic.

What I have yet to discover is whether more web search interest in a particular candidate means that they will receive more votes and ultimately win the election. So I decided to use recent history to see if I could detect whether this was actually an early indicator of future success.

. . .

All of my data seems to point to the obvious conclusion that web popularity may help determine the winner of an election. This thesis has held true until I compared the 2012 Republican primary. The results for this race are very different. The front-runner in web popularity has consistently been Ron Paul.

So like the study by Wellesley College, I have concluded that search results and social media may gauge interest but, when it comes voting, it's still inconclusive as to whether web popularity will ultimately predict the winners."

Another thing I found:

"A group of researchers at Wellesley College examined data from Google Trends and analyzed how effective a tool it could be in predicting U.S. Congressional elections in 2008 and 2010. In highly contested races where data for both candidates were available, the data successfully predicted the outcome in 33.3% of cases in 2008 and 39% in 2010. The authors conclude that, compared to the traditional methods of election forecasting, incumbency and New York Times polls, and even in comparison with random chance, Google Trends did not prove to be a good predictor of either the 2008 or 2010 elections."

Listen, I'm a Trump supporter. I get where you are going, but I seriously doubt that this tool can successfully indicate what you want it to indicate.
I Want To Believe

User ID: 75170101
Australia
08/25/2020 02:13 AM
Report Abusive Post
Re: DEBUNK THIS: Google trends of various elections: Is google more accurate than polls? (It has predicted the winner in all the last 4 elections)
So out of further interest in this, I did some searches on articles on this. The funny part is this article:

How Google searches might be able to predict the 2020 Democratic race

"And that's where this amazing chart -- via Google's Simon Rogers -- comes in. It tracks Google search interest in 23 of the Democratic candidates for president from December 30, 2018, until June 16.

. . .

Now, searching for a candidate on Google isn't the same thing as voting for them. You could, for example, look up "Elizabeth Warren" because you don't want her to win and need some talking points to push back on your pro-Warren friends."

The funny thing is that the chart goes through January 2020 even though the article is dated June 2019, so the chart kept updating apparently even after the article first published.

If you go based on the final date of January 22, 2020, then we should be looking at Bernie Sanders as the current nominee, not Joe Biden. Joe Biden was ranking at 4th on the last date the chart took data.

So based on the above, I took your method of just using last names of Biden and Sanders and ran it first for the same dates as the article and Sanders wins:

I then also ran it until Sanders dropped out. Sanders still is winning:

Here's another article where someone used Google Trends to predict that Romney would beat Obama (something we all now know to not have happened):

Google Predicts Mitt Romney To Win Fall Election?

"My first step was to compare the two candidates for web search Interest within the United States over the last thirty days.

What I discovered is that, according to the Google service, Mitt Romney (in red) is clearly leading in the amount of raw search activity compared to Barack Obama.

. . .

The results show that Romney is currently leading Obama in every State, even States you can assume are "blue" or typically Democratic.

What I have yet to discover is whether more web search interest in a particular candidate means that they will receive more votes and ultimately win the election. So I decided to use recent history to see if I could detect whether this was actually an early indicator of future success.

. . .

All of my data seems to point to the obvious conclusion that web popularity may help determine the winner of an election. This thesis has held true until I compared the 2012 Republican primary. The results for this race are very different. The front-runner in web popularity has consistently been Ron Paul.

So like the study by Wellesley College, I have concluded that search results and social media may gauge interest but, when it comes voting, it's still inconclusive as to whether web popularity will ultimately predict the winners."

Another thing I found:

"A group of researchers at Wellesley College examined data from Google Trends and analyzed how effective a tool it could be in predicting U.S. Congressional elections in 2008 and 2010. In highly contested races where data for both candidates were available, the data successfully predicted the outcome in 33.3% of cases in 2008 and 39% in 2010. The authors conclude that, compared to the traditional methods of election forecasting, incumbency and New York Times polls, and even in comparison with random chance, Google Trends did not prove to be a good predictor of either the 2008 or 2010 elections."

Listen, I'm a Trump supporter. I get where you are going, but I seriously doubt that this tool can successfully indicate what you want it to indicate.
Quoting: TrustNoOneKS

Look at the OP ...and use the method described
I'm a J & proud zio.

FE Challenge: Provide a formula which calculates the exact distance between 2 GPS coordinates that does not use the Earth's radius of 6,371 km in it's assumptions

Weyoun

User ID: 78625468
United States
08/25/2020 04:25 AM
Report Abusive Post
Re: DEBUNK THIS: Google trends of various elections: Is google more accurate than polls? (It has predicted the winner in all the last 4 elections)
Weyoun

User ID: 78625468
United States
08/25/2020 04:31 AM
Report Abusive Post
Re: DEBUNK THIS: Google trends of various elections: Is google more accurate than polls? (It has predicted the winner in all the last 4 elections)
California would be vast majority Trump except for election fraud. And this year I really hope to see a intense scrutiny and secure votes. Because Dems are behind both covid and the riots. They're so fucking evil its unbelievable what we're living through.
Quoting: Sungaze_At_Dawn

Commifornia is mailing ballots to the entire population of TIAJUANA, including all the gravestones and pets, and pets' gravestones.

Weyoun

User ID: 78625468
United States
08/25/2020 04:39 AM
Report Abusive Post