The mystery behind a group of electoral maps

by Eszter Hargittai on October 23, 2016

Yesterday, when I was going through my Facebook feed, I saw several people in my network post a copy of the map below. (As far as I could tell, 9-10 people in my network had shared it, I figure this from the fact that I saw two and then saw a link “8 shares” or “8 more shares” below them.) To clarify, the images I saw were posting the map without the question that I overlaid on it. I am not posting the original so as not to perpetuate what I think is likely misinformation circulating. As a point of comparison, it is very rare that that many people in my FB network post the same thing, or at least FB doesn’t seem to suggest it often. Three of the people who posted it were academics, one works in the policy realm, all work on Internet-related topics. I mention that simply to note that people of all sorts may be prone to spreading online what seems like factual information without necessarily knowing its source. (See below for more on why I don’t know who the other people were, a bit of a mystery in and of itself.)

Seeking source for group of electoral maps

As far as I can tell, there is no source listed on the image. My searching led to all mentions of it linking to the same image-sharing site, one that as far as I know is associated with people sharing images on Twitter. There are lots of mentions of it on Twitter. But scrolling all the way to the end doesn’t clarify (not on my list of results anyway) who may have been the person to share it on Twitter first since the first link I see actually uses a Facebook short URL. (I guess that could have been the person, but there is nothing to confirm it. There seem to be all sorts of FB links on that person’s Twitter feed that no longer exist.)

What has fascinated me about this map is the life people have given it on social media. (To the credit of mainstream media, I have not seen any mentions of it in such outlets.) People have not simply reposted the image, but have added commentary in several cases taking its substance for granted. The maps look a lot like what one can see on Fivethirtyeight.com, but Fivethirtyeight puts its logo on its maps, which you can see when looking at its images. (If you compare the maps carefully, some details are also different from what Fivethirtyeight does.) Of course, there could be other reputable sources. But this image lists no sources.

Now moving on to what I consider another mystery. After commenting on two of my friends’ posts on Facebook asking whether they knew the source for the image (not to be difficult, but because I am truly curious), I wanted to check the mentions of the other people who had posted it. I remembered seeing two more names and I could still see it when going directly to their Walls. But I could no longer find any of the shares in my feed. So something that Facebook had thought important enough to show me on the top of my feed a few minutes earlier was no longer showing up at all. I scrolled and scrolled and scrolled and nothing. Then I decided to rearrange the newsfeed chronologically (what FB calls “Most Recent”, instead of its default “Top Stories” ordering). I scrolled and scrolled and scrolled again well beyond the time when I knew one of the people in my network had posted it and still nothing. This suggests that the “Most Recent” listing of FB posts also doesn’t show all posts. The only way to track down a post for sure seems to be to know who put it up and go to their Wall. FB is constantly rearranging your feed and it turns out that if you don’t catch exactly who had posted something, you may never be able to track it down. (I realize that sounds very dramatic, but in a case like this it would have been nice to be able to do so.) In case you’re thinking that they didn’t show it to me, because they figured they already had, I guarantee you that is not it. FB shows me posts I had already seen all the time, too often if you ask me. So that can’t be the reason.

I am very good at Internet searches, but I am sure I don’t know all the tricks. If someone can track down the origins of that image, please share. Every mention I have been able to locate points back to the same pbs.twimg.com/etc URL and mentions no source.

{ 73 comments }

1

Rajiv Sethi 10.23.16 at 9:06 am

2

Eszter Hargittai 10.23.16 at 9:08 am

Yes, they are copies of them. (They are not the same images.) I have seen those posted all over. But that doesn’t explain the rest and most people are commenting on the others when they post this image.

3

Rajiv Sethi 10.23.16 at 9:34 am

I think the image you’re referencing originated on twitter since it’s stored at:

https://pbs.twimg.com/media/CvMZKsMWIAAVd8M.jpg:large

which is a domain used by twitter to store user-uploaded images. So someone created and uploaded the image (inspired by the 538 post I presume) and it then went viral. Not sure who that someone is but it’s possible to find out using Microsoft’s ViralSearch. See Jake Hoffman’s video here:

http://searchengineland.com/microsofts-viralsearch-search-engine-for-measuring-tweets-viral-content-150703

But this tool is not publicly available.

4

Foppe 10.23.16 at 9:37 am

Apologies for going meta, but I’m having rather a hard time understanding why these images are interesting. Even if only people of color and women voted, the Dems, given that they only represent professional class interests (see Tom Frank’s Listen, Liberal), would still ignore their interests, because of pols’ conception of politics, “how the world works”, etc. So why does this matter, absent a change in mental conceptions?
As I see it, all this is is a visualization of the Democratic Party dream that demography will win them the future. And on that topic, they are quite mistaken, as they are wholly ignoring the effect their not meaningfully representing voters has on turnout rates. For an illustration of this, take the British turnout during the referendum: it was as high as the turnout in the election in which the voters got rid of the Tories in ’92, after which every (re)election saw turnout rates that were between 10-20% lower, because people caught on to the fact that voting had become useless now that the “left” party had stopped representing lower/working class interests in favor of the highly educated / affluent professionals who politicians associate with, envy, and view as their peers / betters.)

5

Eszter Hargittai 10.23.16 at 9:46 am

Rajiv, classic mansplaining behavior. You are repeating what I had already said in the post by pointing to the URL I had already linked to and mentioning that it’s hosted by Twitter, something I had already stated. “Not sure who that someone is” was precisely what I was asking about and it looks like you don’t have an answer.

Foppe, I see no point in debating the substance of the maps given that we don’t even know if they are for real since there doesn’t seem to be a source for them.

6

nastywoman 10.23.16 at 10:08 am

‘Foppe, I see no point in debating the substance of the maps given that we don’t even know if they are for real since there doesn’t seem to be a source for them.’

The maps are ‘for real’ – as much as they just seem to visualize some data of 538 – the same way 538 build their maps?

And I thought that the ‘substance’ of just the 538 maps were fascinating – as they show that WE -(the nasty women) – will decide the US erection.

7

RichardM 10.23.16 at 10:11 am

The graphics template comes from http://www.270towin.com. That site allows you to click on states, mark them red, blue or brown, and share the result. It doesn’t have any feature to feed in polls and get a result, so if anyone did that, they did it offline.

Given most polls have Trump losing minorities 99/1, that map for is presumably accurate. The others could well be from someone pressing states at random until they got a result that felt right.

8

Rajiv Sethi 10.23.16 at 10:12 am

Sorry for causing offense, I totally missed the url at the bottom of your post and took it from my own FB feed. Got curious and carried away. My mistake, should have been more careful.

9

John Quiggin 10.23.16 at 10:23 am

Thanks for raising this, Eszter

I’ve seen these maps too, and wondered about the source, but didn’t get around to checking it. It sounds like it would have been a thankless task to do so.

10

ZM 10.23.16 at 10:35 am

The earliest I could find the maps appearing with a Google search was on this blog 2 days ago in a post https://heartiste.wordpress.com/2016/10/20/tribal-electorate/

It doesn’t say if the bloggers created the maps or if they found the maps elsewhere, I asked them about it.

They wrote on their post using the maps that the maps show POC are less ideologically diverse than white people, and said that the maps show that “a big reason the Verbal Class is getting outgunned in social media is because they are deficient in visual-based mediums that emphasize the transmission of gut-punch memes and neutralize the sophist’s advantage of beguiling foes with an avalanche of wordy evasiveness.”

I told them there was a Crooked Timber post on the maps questioning the source and accuracy.

11

Haftime 10.23.16 at 10:40 am

Don’t know who made it but this appears to where the ‘data’ comes from.
http://projects.fivethirtyeight.com/2016-swing-the-election/

You can reproduce the figures in the image using the turnout sliders.

12

Rajiv Sethi 10.23.16 at 10:41 am

Found a couple of non-twitter sources from 10/20 by googling “The reason I didn’t break down the POC vote” but reluctant to link because of some nasty racist language.

13

ZM 10.23.16 at 10:43 am

This is discussion board conversation about it here too http://www.lipstickalley.com/showthread.php/1097959-What-the-Electoral-Map-would-look-if-only-these-people-voted

It is unsourced and when asked someone else gives the source of Nate Silver’s original 2 images about women and men voting differently that turned into an internet meme.

I am not going to quote people on this discussion board, but they are discussing the racial implications as well, like the blog I mentioned above. In contrast to the blog some people on this discussion board are saying Hilary Clinton would be toast without black people, and white women aren’t voting for her. Some people have questioned the accuracy.

I would guess from these two pages that someone has made the image for the racial implications. This is what people are picking up on, on the blog and on the discussion board.

14

Eszter Hargittai 10.23.16 at 10:46 am

Nastywoman, other than the two gender-only maps as noted by Rajiv already above, we still don’t seem to have a source for the other data. Can you point to a source? (I don’t mean generally invoking 538, but the actual data or a map with the same info on 538.)

RichardM, very helpful, that does indeed look like the source of the graphics and suggests that random or data-free-purposeful clicking could be at play to display some stories someone wanted to display.

ZM, interesting, that seems to be the earliest mention of the image so far, most of the references I’ve seen have been on the 21st. Thanks.

15

Eszter Hargittai 10.23.16 at 10:53 am

Haftime, so if you slide to 100% turnout vote? That’s likely.

When I say source, by the way, I mean the data plus who actually created the image and posted it.

Thanks, Rajiv, yeah, no need to link to outright racist pages, we can each follow up on the search if we want. I think ZM is right that there was probably racial motivation in creating these maps.

We can easily find a source for the gender-only maps, but anyone has yet to point to a straight-forward source for the racial breakdowns.

16

Collin Street 10.23.16 at 10:57 am

Reverse-google-image-search [where you give google a file and it tells you about it] gives the result “Best guess for this image: anti breastfeeding in public”

Which… tracking it down gives me a huffPo article on breastfeeding that uses what appears to be a copy of the same master map…

http://www.huffingtonpost.com.au/entry/map-where-its-legal-to-breastfeed-in-public_n_5637301

Which suggests you might find something on huffpo, but I don’t know that site at all.

17

Collin Street 10.23.16 at 11:04 am

… but the huffpo graphic doesn’t have the state labels. Hrm. Also, the black text font is Arial, which suggests that it might not have been done by a pro graphic designer.

18

Haftime 10.23.16 at 11:07 am

Sorry, I’ll explain better. You get the identical results to the ‘mystery picture’ (including the race + college) if you set every turnout slider except for the relevant group to zero (100% turnout is different – because turnout is defined at fraction of people w/in a group that vote, not % of voting population from a given group). I have not checked every state for every map shown, but the overall figures are the same which, in combination with the fact that only and all the maps included in the mystery figure are ones derived from that page makes me believe that this was the process done by the anonymous author.

I’m afraid I can’t help anymore on the author of the figure – I suspect it will be hard to tell because both the map generator and data are public. Good luck. At least it is possible to assess the reliability of the data now: these maps are as reliable as 538.

19

Eszter Hargittai 10.23.16 at 11:08 am

Collin Street, you just pointed us to a post from 2014, which is unlikely to be the source for the map in question.

I am familiar with reverse image searches and various other modes of searching the Web. I didn’t post things that didn’t work out. Please don’t post links to places that *don’t* have information about the map, that leaves a lot of options (like hundreds of millions).

20

Collin Street 10.23.16 at 11:15 am

Just saw RichardM’s; the labels on that graphic are in Helvetica [G in Georgia, R in Oregon], or at least look like that on this computer.

So we’re building a picture of the pictures being labelled — probably in MS paint or equivalent — by someone who used the site RichardM linked to. Could be anyone.

21

Eszter Hargittai 10.23.16 at 11:15 am

Thanks, Haftime. You’re right that I confused the relevance of the 100% turnout here (not relevant here). So yes, this is a plausible source. It would take a bit of work to check all 50 states for all six maps, but it looks like we have a lead. Thank you.

Still curious that the person who posted it didn’t add source (either of the data or their own name/affiliation).

22

Collin Street 10.23.16 at 11:23 am

Collin Street, you just pointed us to a post from 2014, which is unlikely to be the source for the map in question.

Not the map per se, the background clipart. You build libraries of these sorts of things, for the stuff you use a lot of; different organs will have slightly different libraries, so if you recognise a house-internal master graphic you can identify the organ that originated the finished graphic. The detail positioning of the big island of hawaii next to the southern bit of texas, say, looked like a match to me.

Was my thinking. “It might have come from XX, it fits their graphic style”. Turns out I was wrong, of course.

23

Collin Street 10.23.16 at 11:24 am

[also it was a comment on the uselessness of reverse-google-search, of course; on the face of it breastfeeding is clearly not relevant to the topic at hand…]

24

Eszter Hargittai 10.23.16 at 11:29 am

I’ve been playing around with that 538 page (see Haftime’s post above) and Haftime’s suggestions seems right. If you leave the one category of interest at the default of the site (don’t move the dot) and then move all others to zero then you get these maps. Note that it is extremely sensitive to the assumption of the placement of that dot. If you move that dot even in the slightest, states start moving from one side to the other. Also, they are in very different shades, but in the maps on this image, they seem to be uniform.

Now that we’ve figured out this part (thanks, Haftime!), any comments on the mystery of FB’s feed? This is just one of a few examples I’ve noted recently.

25

Eszter Hargittai 10.23.16 at 11:30 am

Collin Street, sorry, I missed the snark about how useful – or not – these search tools are.

26

nastywoman 10.23.16 at 11:30 am

‘Can You point to the source’

As these maps in question seem to include some kind of ‘Do it yourself Maps’ – based on the breakdown numbers of 538 – anybody on the internet might be able to be ‘the source’?

Or not?

27

Eszter Hargittai 10.23.16 at 11:31 am

Nastywoman, yes, then we’re on the same page: anybody. So the question then is: why do people just trust anybody?

28

nastywoman 10.23.16 at 11:35 am

‘why do people just trust anybody?’

The internet doesn’t trust anybody – anybody gets questioned?

29

Eszter Hargittai 10.23.16 at 11:37 am

Here, yes. :) But the posts I was talking about and many Twitter shares very much took it for granted.

30

ZM 10.23.16 at 11:43 am

Haftime,

“I have not checked every state for every map shown, but the overall figures are the same which, in combination with the fact that only and all the maps included in the mystery figure are ones derived from that page makes me believe that this was the process done by the anonymous author.”

If you look at the graphics 538 have, the images on this map seem to me to underestimate white women, white college educated, and white non-college educated, and white men support for the Democrats.

I think its interesting that they underestimate white non-college educated support for the Democrats so much, its about 3 states on this map, but looking at 538 its about 12 states and a couple on the border of 50% support for Trump.

Now I really better go do something else.

31

nastywoman 10.23.16 at 11:47 am

‘But the posts I was talking about and many Twitter shares very much took it for granted.’

Did they take it as much for granted as Trump Twitters?

And I don’t mean to be silly – but I don’t have FB – as most of my friends who have FB take most what ‘our’ friends are sharing on FB or Twitter ‘for granted’ – and what did Groucho Marx so famously say: “We shouldn’t care to belong to any club that will have US as a member’.
(Silly!!)

32

Eszter Hargittai 10.23.16 at 11:50 am

ZM, yeah, sorry, this can be a complete time suck. As I note above, the results of the map are *extremely* sensitive to the location of the dot. Move it by just one percentage point on either axis and you easily get a different result.

33

Eszter Hargittai 10.23.16 at 11:51 am

as most of my friends who have FB take most what ‘our’ friends are sharing on FB or Twitter ‘for granted’

This being an example of a larger trend doesn’t excuse it though. ;-)

34

RichardM 10.23.16 at 11:54 am

People trust ‘anyone’ because what was the work of one person in perhaps 30 minutes looks like it was the work of an organisation with separate research and art departments. 5 years ago it probably would have been, 15 years ago it couldn’t have been anything else.

35

nastywoman 10.23.16 at 12:00 pm

‘This being an example of a larger trend doesn’t excuse it though. ;-)’

It doesn’t at all –
but as this might be also a trend of more ‘Do it Yourself Graphics’ -(and other ‘content’) – on the internet – some ‘self sourcing’ could become pretty… interesting?
-(if that’s the right word?)

36

marcel proust 10.23.16 at 12:36 pm

Collin Street wrote:

… but the huffpo graphic doesn’t have the state labels. Hrm. Also, the black text font is Arial, which suggests that it might not have been done by a pro graphic designer.

But what about the kerning, eh?

37

Lee A. Arnold 10.23.16 at 12:40 pm

This map image swept through a dozen alt-right nationalist/racist blogs and forums on Oct. 20-21. It may have been created by one of them which is being coy about its origin.

This may be in response to a Fivethirtyeight story on Oct. 11 showing that women are breaking for Hillary:
http://fivethirtyeight.com/features/election-update-women-are-defeating-donald-trump/

The racists clearly have taken objection to the implication that WHITE women are abandoning The Donald — because the larger map-set is being used at some of the sites to argue that the 538 story is false because it does not account for race.

Whatever the provenance, even an innocent one, it is an erroneous bit of wish-projection. Aside from secret internal polling by the campaigns themselves, no one is likely to have detailed information at the present.

Indications from an assortment of public polling are that “College Educated Whites” and “White Women” are breaking for Hillary in several of the swing states. Several pollsters suspect a big increase in the “sure, honey” factor that has occurred in every election, in which one spouse doesn’t know how the other spouse will vote.

38

Eszter Hargittai 10.23.16 at 1:44 pm

Here’s an interesting response I received. On Facebook, the image shows pbs.twimg.com in the lower-left corner as that is the site where it’s hosted. Someone responded to my query with this: “It says at the top that it’s from PBS if that helps.” The site has nothing to do with PBS, the letters “pbs” are in the URL, that’s it. I had wondered earlier whether anyone would mistake this for PBS. Now I know.

39

P O'Neill 10.23.16 at 1:52 pm

On the broader context for the query, this from New York Times magazine a couple of months ago on Facebook political culture is well worth a read.

40

F 10.23.16 at 1:55 pm

The maps themselves look like they might have been created at 270towin.com. The bars on top don’t look right though.

41

Neville Morley 10.23.16 at 4:03 pm

I would be interested in further discussion of the FB timeline thing, from people who have a better idea how these algorithms work. I’ve got used to it picking and choosing which posts from friends appear, but realising that it was then revising the past as well came as a nasty surprise. Basic assumption is that users have butterfly minds and demand constant novelty and stimulation?

42

Neville Morley 10.23.16 at 4:04 pm

From the perspective of a historian, I find it disturbing that the past can be so easily rewritten. What am I supposed to do, screenshot everything?

43

nick j 10.23.16 at 5:29 pm

44

Eszter Hargittai 10.23.16 at 5:32 pm

Neville Morley, indeed, tricky for sure. I actually do screenshot a bunch of things, but those are just the things that I have already seen and know that I may want to revisit later. There’s little you can do for content you are only encountering later. There is the Web Archive, of course, but that doesn’t archive things at the level of private (semi-private) social media accounts.

No one other than some people at FB really know what’s going on with their algorithms. There have been some studies, well, just a few so far, looking at this empirically. And then there are people’s personal experiences. (I have another one that I want to write about, but too long for a comment.) But generally speaking, we the masses don’t have much to go on in this realm.

nick j, it doesn’t seem to help, none of those seem to add new info.

45

Martin Bento 10.23.16 at 5:49 pm

The 538 post someone linked is almost 2 weeks old, before the second debate and not necessarily fully accounting for the impact of the 2005 tape. The new ABC tracking poll has Clinton up by 3 points among men.

http://abcnews.go.com/Politics/clinton-vaults-double-digit-lead-boosted-broad-disapproval/story?id=42993821

That may be an outlier, but Romney won men by 20 points or so.

46

Martin Bento 10.23.16 at 6:14 pm

Sorry, that 20 points was hearsay and evidently confused. According to the Roper Center, Obama carried women by 20 points and Romney carried men by 8.

http://ropercenter.cornell.edu/polls/us-elections/how-groups-voted/how-groups-voted-2012/

What’s really interesting is that the ABC poll I cited above, one of the most favorable for Clinton, also has Clinton 20 points ahead among women. The difference is that she is doing much better among men. That would be very surprising for the first viable female candidate even without all the scandal about Trump’s words and behavior. Even given all that, Clinton is only matching Obama among women. She’s just brining along a lot more of the “bros”. Not at all the narrative I’m hearing.

47

Martin Bento 10.23.16 at 7:16 pm

I should have my coffee before I comment. I misread that. Delete previous comment and this if you like.

48

Martin Bento 10.23.16 at 8:37 pm

OK, since you kept it in, the point stands. The number I cited was a (slightly-rounded) measure of the actual gender gap in 2012. Obama carried women by 11 points, and Romney carried men by 7, resulting in a gender gap of 18.

The LA Times poll has Clinton up 20 among women and up 3 among men, gender gap 17.

Here are some other recent polls and the gender gaps:

CBS news, October 17, Clinton up 19 among women, down 2 among men, gender gap 21

http://www.cbsnews.com/news/cbs-poll-clintons-lead-over-trump-widens-with-three-weeks-to-go/

Investor’s Business Daily – Oct 23 Clinton up 8 among women, down 12 among men, gender gap -20. (I’m counting a gender gap that favors men as negative, just to flag it. Yes, this poll currently shows Trump slightly ahead, unlike most)

http://www.investors.com/politics/ibd-tipp-presidential-election-poll/

LA Times – October 23 Clinton up 9.9 among women, down 11 among men, gender gap -.20.9. Also, Trump advantage.

http://graphics.latimes.com/usc-presidential-poll-dashboard/

Quinnipiac – October 19 Clinton up 11 among women, down among men 6, gender gap 17

https://www.qu.edu/images/polling/us/us10192016_U29frgv.pdf

The polls most favorable to Clinton show a gender gap about the same as Obama’s. The polls unfavorable show a comparable gap, but with absolute numbers more favorable to Trump. If there is any election where one would expect a gender gap to widen, it would have to be one between the first viable female candidate and a proud groper. But it doesn’t seem to be happening, even though the groping issue does seem to be helping the female candidate. Why?

49

Martin Bento 10.23.16 at 9:00 pm

In the first paragraph above, I meant ABC poll, not LAT. Link is in comment above.

50

Kirk Lambert 10.23.16 at 9:49 pm

Does this help? The map uses 538’s data, extrapolated from 2012.

https://medium.com/@ste.kinneyfields/do-you-know-this-graphic-i-made-it-heres-why-f97bcf88408c

51

Tomsk 10.23.16 at 10:11 pm

Foppe @ 4 – just to note that UK voters did not in fact ‘get rid of the Tories’ in 1992.

52

adam.smith 10.23.16 at 10:44 pm

@Kirk — ah. I had seen thihs but couldn’t find it again, thanks for posting. Assuming the poster isn’t falsely claiming credit, I think that solves provenance (and methods). It doesn’t quite explain why it went viral, but that’s always hard to answer.

53

Kiwanda 10.23.16 at 10:51 pm

Mystery solved by Kirk Lambert, sounds like, and the maps were made pretty much how people here figured.

Although I kept thinking until a minute ago that the 538 page was using Clinton/Trump polling data, the page clearly says (as Kirk Lambert notes) that it’s based on Obama/Romney data, I think, adjusted for demographic trends. (Maybe somebody said that before, but I didn’t notice.) So it’s about the 2016 presidential election only in a limited way.

54

Doug O'Keefe 10.23.16 at 11:55 pm

So, to get to the heart of the thing for dummies like me, does the method of measuring by assuming 100% or 0% turnout exaggerate the extent of American white men’s sympathy with the Republican party?

(Coming from California, my state’s results don’t seem unlikely: it says most non-college-educated white men support the Republicans, while the majority of all the other groups goes Democratic.)

Maybe I missed a post where this was addressed?

55

Sara R 10.23.16 at 11:57 pm

56

Erin Fields 10.24.16 at 2:15 am

Was just about to post the link from the commenter just above mine. This was a post-for-friends-that-somehow-went-viral, which is why there wasn’t citation on it. More explained in that blog post.

There is also an updated map with Quinnipiac poll from 10/19:
https://medium.com/@ste.kinneyfields/updated-electoral-college-image-baba73948592#.e17lc477m

57

ZM 10.24.16 at 4:00 am

I thought Facebook’s timeline newsfeed thing runs like a search on your Facebook likes and friends. When you like new things it changes, or as other people like things. I kind of think of it like a river and just save things I want to look at again some other time.

58

ZM 10.24.16 at 4:33 am

The blog post by the author linked to above is good. I had my doubts about the data being accurate, it’s a real eye opener seeing that level of racial and educational disparity in the election voting intentions. We’ve had a few discussions on CT where race comes up as a factor in this election, and the map now it’s shown to be proper, really does visualize the racial issue very well.

59

pnee 10.24.16 at 5:06 am

Tangentially, I wish people would use the maps with state area adjusted for electoral vote strength for this sort of thing. 538 does both, but they still headline the traditional geographic map, unfortunately. Scroll down on their main presidential forecast page and see the one made of hexagons, if you don’t know what I mean.

60

Hotel de Phil 10.24.16 at 5:53 am

I was wondering the same thing myself, so thank you for posting this.

Now that we know where the data is from, it seems that these maps don’t really say anything about this election, which is Trump v Hillary. As described in 538’s post, It’s just projecting Republican/Democrat votes from the 2012 election to this year’s election, and adjusting for demographic changes. It does not account for the Trump or Clinton factors. It assumes the groups who voted Republican or Democrat in 2012 would vote the same way in 2016, but we know that’s not the case. In some states Hillary does better and in some states Trump does better.

The updated maps (based on a recent poll) in the link in Erin Fiels comment above show different results.
https://medium.com/@ste.kinneyfields/updated-electoral-college-image-baba73948592#.iqykwhkw6

61

Eszter Hargittai 10.24.16 at 6:20 am

Kirk Lambert and others, regarding that link: that account on Medium has no activity from before 24 hours ago (well after the map initially circulated). Where is the original post that put out the image? Who is to say someone who had nothing to do with the original map didn’t just create that post with information available on a thread like this? The Medium post says “I moved some toggles, plugged my results into 270towin, posted, ..”. Posted where? Where is the link? There are other links to sources in the Medium post, but not to the post where it all started. That seems curious to me.

Medium accounts are often associated with Twitter accounts of the same name, but there is no Twitter account with that handle either.

62

ZM 10.24.16 at 8:09 am

Eszter, the author of the post Ste Kinney-Fields has the image on a Facebook post on 17th October. This is probably where it started circulating through Facebook from I guess.

63

engels 10.24.16 at 8:30 am

Where is the original post that put out the image

Maybe I’m missing the point but I think Erin was claiming the original post was private and only accessible to friends (perhaps on Facebook).

64

Eszter Hargittai 10.24.16 at 8:48 am

Is there a mention of a private Facebook post in the Medium post? I didn’t see one. Nor a reference to an Oct 17th posting. Nor a screen shot of the private post. I’m not saying that it didn’t happen, I’m saying that there is no evidence of it in the post as far as I can tell.

65

ZM 10.24.16 at 10:02 am

If you search for the author on Facebook you can see his public posts and its a public post from 17th October. He would have had to steal the image and claim it was his on Facebook on the 17th October if he isn’t the genuine author. He is really apologetic in his article about not providing sources, and also about not putting explicitly anti-racist text on the image itself.

66

Eszter Hargittai 10.24.16 at 10:11 am

ZM – engels suggested that the post was private, not sure based on what. I did search as you suggested and see it now. It is public so one can link to it:
https://www.facebook.com/photo.php?fbid=10157627878010615&set=a.474813550614.379228.514860614&type=3

And you can see the edit history as well by clicking on the little down-arrow symbol in the upper-right corner of the post and choosing View edit history (or “More Options” if that doesn’t show up immediately). It was first posted on Oct 17th at 6:38pm updated a minute later to add info about data and graphic source (without details about the parsing of the data, but people above have already figured that out).

67

Saurs 10.24.16 at 11:33 am

(Ste Kinney-Fields doesn’t go by male pronouns, by the by.)

68

ZM 10.24.16 at 12:06 pm

Saurs,

Oh, thank you, I didn’t realise, I was thinking it was short for Stephen or something I never heard the name Ste before

69

RINO economist 10.24.16 at 12:43 pm

There is an odd juxtaposition of rhetoric in the source-post of the statistics

https://medium.com/@ste.kinneyfields/do-you-know-this-graphic-i-made-it-heres-why-f97bcf88408c#.ycumf0105

“If only white women voted … [sic] then white women would still have a lot of work to do confronting their internalised racism, sexism, Islamophobia, anti-semitism, etc. All of them. Even the ones voting “Blue” or Green” in this election.”

followed in the text soon after by

“something all white people should be held responsible for.”

versus

“My original point was that white men are not the only people who are responsible for Trumpism.
Likewise, and I can’t believe I need to spell this out, not all “white men” are directly responsible for Trumpism. Trump, Trump supporters, and the Republican party that created the environment that allowed him to rise are the ones responsible for Trumpism, regardless of their demographics. But the demographics are still notable and in keeping with Trump’s rhetoric.”

The word “directly” in “directly responsible” seems to be doing a lot of heavy lifting here. I find the first set of quotes pretty offensive, but that may just be me.

70

engels 10.24.16 at 12:53 pm

Ah—that makes sense (I was conjecturing based on Erin Fields’ description @56: ‘a post-for-friends-that-somehow-went-viral’.)

71

Ragweed 10.24.16 at 4:31 pm

One other factor with finding things on Facebook is that I believe that posts that one of your Friends “like” will show up in your newsfeed, but will not necessarily show up on the Friend’s timeline. Usually posts that are shared and posts where one is tagged show up on the timeline, but not posts that are just liked (though there probably is a setting for that). So even knowing which of your Friends were responsible for something ending up on your newsfeed does not mean you can go back to that Friend’s page and find it again.

72

hix 10.24.16 at 5:08 pm

Maybe facebooks algorithm did decide you were being difficult in a non facebook usetime maximicing way. Trying to understand facebooks manipulations is sort of pointless. What seems clear is that there is no reliable easy to understand public rule…

73

Ethan 10.24.16 at 10:15 pm

If I read this correctly, the only thing hard-coded into the tool on five-thirty-eight is demographic shares of electorate. The default values are from the 2012 election, but in addition to changing turnout, you can also change support levels by moving the dots (you can also change the third party support level). So, to get Clinton/Trump numbers you would have to use a current poll with a demographic breakdown or an average of several to decide on a Trump/Clinton/Third party for each subdivision. Then you would have to drag the dot for your group to the right support point (which isn’t the percentage voting R or D, but the percentage of the non-3rd party voters voting R or D) and then reduce all other groups to zero turnout. As long as you’re doing one group at a time, the turnout of that group can be anything. However to calculate a total “White people” result, you also need to make decisions about the relative turnout levels of the two groups, which means you need to decide which you have more faith in calculating, the turnout of white men vs. white women or the turnout of white college vs. non college voters and choose on of those classification splits and adjust those categories to what you think they should be. Not sure if anyone has even attempted to do that, it would be fairly difficult and give plenty of opportunities for bias to influence the simulation (both the bias built into the original polls and the bias of the aggregating models assumptions).

Comments on this entry are closed.