How the popular vote was decided - numbers!

[seaclyr]

I've been exploring, with much help from jbond56, the numbers from cnn that he made available. We first looked at how the percentages for each candidate changed over election night and found a surprisingly large early lead, percentage-wise, for Bush which dropped off over election night.



We next looked at actual vote differences, which show two discrete waves of votes that favored Bush. Each totaled about 1.8 million. The first wave came in early on, beginning as far as we can tell when votes were first reported, and was over by the time about 20% of the total vote was in. The second came in significantly later, beginning at around 65% or so of the total vote and was over when about 85% was in.

Several interesting points emerge from this:
1. It may (or may not?) be coincidence, but 2 x 1.8 million votes are more than enough to provide the final Bush margin.
2. It is remarkable how rarely the Bush lead in the popular vote dips in the favor of Kerry; when it does dip, the change is typically very small until very late in the evening when more than 90% of the vote was in.
3. With the exception of the pro-Bush waves in the vote-difference plot there is remarkably little variation seen over election night as total vote counts show. The fact that in general there is little variation seems odd, as does the fact that what variation there is seems oddly systematic.

Finally, a word of caution in interpreting these numbers: they need to be compared with other elections - we are working on getting numbers for 2000.

[truehawk]

1. Compare these to the Gallup Numbers from 2 weeks before the election

I was worried for months that they were laying the groundwork for fraud by having Gallup oversample Rethuglicans.

[whistle]

2. You appear very astute in working with numbers and statistics...

...It is a great analysis. Now, would you have any way to run a similar analysis of the Gore vs Bush numbers for the 2000 election? My thinking here is that the 2000 election was probably the closest presidential election in modern history and a time phase comparison such as you did above, could reveal the irregularities you are suggesting throughout the course of the reporting of results following the close of the various county and state precincts.

Then, if is is possible to run the two steams of results up until the beginning of the Florida recounts, you would then run a "test of the significance of the differences" I believe it's called to see how unusual what you describe above really is. Perhaps the same thing could be run on prior elections going all the way back to Bush I and his victory in 1988. Keep up the great work!

[seaclyr]

6. Yes, Gore vs Bush is hopefully next

The plan is to do a similar analysis for 2000, looking for irregularities late in the process. If it's there, it really could show up.

[jamboi]

3. Very interesting! n/t

[understandinglife]

4. Superb analysis; may I send a link...

...to a few folk before you finish your comparison to the 2000 national election (that you mention in your final sentence)?

Peace.

"Did Bush Know?"

[seaclyr]

8. Yes, I'll pm you the link

Sorry, your pm seems to be disabled. If you contact me I'll provide the link.

[understandinglife]

9. Thank you. Just set my 'pm' so I can receive them. (n/t)

[Hamoth]

5. States reporting make the dif?

Is it possible that these surges are the results of when certain states reported in their initial tallies? I mean, were they reporting reds then blues then reds again?

Hmmph...actually...it should be the opposite.

[seaclyr]

7. My thought too...

intuitively it should be the opposite, right? (It's actually what got me thinking along these lines). But intuition isn't always right, which is why results from 2000 are essential.

[Zan_of_Texas]

18. Yeah, it should be the opposite --

If states reported in waves by time zone, the East Coast and West Coast should have been Kerry, and the middle Bush -- but the graph shows Bush increases more sharply at the beginning and end of the reporting.

[breadbox]

21. early blues

Unfortunately a lot of early states are blues (like South Carolina,
Virginia, perhaps NC,....) In fact, while watching the returns I
was prepared for the initial 30 minutes of depression while the
early returns came in --- after all, NY didn't close for what, 2 hours?,
after SC.

It was the second wave of bush votes that sparked the month long
depression since......

[RaulVB]

10. Great analysis

Have you send these graphs to Cliff Arnebeck?

Regards.

[seaclyr]

20. I emailed him tonight

with a commentary and links to the graphs

[liberal43110]

11. Think of this vs. exit polls

If I understand this correctly, it is even more interesting in light of the exit polls showing strong Kerry leads in many places all over the country. One "explanation" (a wrong one, in my opinion) in the exit poll discrepancy was that Kerry voters came out early--so we may have expected to at least see some actual vote counts in favor of Kerry...but not according to your work. Hmmmmm.....

[Emillereid]

12. What states were reporting in when the two jumps in Bush's totals

happened. My recollection was that several extremely red states reported in early which gave Bush his initial lead. What states had come in when the second jump occurred -- could it have been other very red states. The pattern I remember from watching the results come in Nov. 2 was that the red states went overwhelmingly for Bush -- first the South fell and then the mid-west and western red states -- both in a clump as the polls closed. Could this account for your data?

[jbond56]

14. I'm looking at the video now.

I'll work on extracting which states are reporting. By the second spike the west coast was reporting numbers.

Now for something odd.

About the time of the first spike is when they cut to the whitehouse. Around the second is when they started reporting victory signs at the white house.

[Emillereid]

15. Very interesting ... I remember at the time the WH shot with

the family stunk to high heaven. The polls were still open here in CA -- and I thought it was just free advertising for him.

[Straight Shooter]

16. Nothing "odd" about it.

Nope, not a darn thing odd about it, once you factor Rove in the equation.

I read of the impending Monkey Palace conference, and my first thought was, "The fix is in."

You might call it intuition. I think it's just foreknowledge of bush and how he and his minions operate.

Edit: Pardon my manners. I forgot to mention that's an excellent analysis. :thumbsup:

If you don't have Cliff Arnebeck's email handy, it's [e-mail]

[seaclyr]

19. Thanks for the email address!

[seaclyr]

17. The popular vote is a running total as I understand it

I believe numbers came in before a state "fell" one way or another. For example the first numbers reported by cnn were in the tens of thousands, I'm guessing not enough for a state to be called for Bush. It's a good question where early numbers reported by cnn (via Associated Press of course) came from. So I don't yet know the answer to your question, but we may be able to figure it out.

[m.standridge]

13. e.g.: Kerry led Bush at least once in AR polls and this was BEFORE Clinton

began campaigning for him.
I'm not one to "call" AR for Dems in past elections, don't know it did now, either.
But exit polling here missed something this time, I could feel it.
This is aside from any tampering vis a vis the exit poll discrepancy articles in academic circles at Univ Penn.

This was last-minute Clinton. He bridged a major-league gap among Demos that may have hurt Kerry some nationally--unenthusiastic female voters who tended to be "unlikel" voters over Vietnam. With Clinton on board here, this gap tended to disappear in final few hours.
Statewide media here--e.g., KTHV Channel 11, a CBS affiiliate, stated on election day that polls here were then "48-48". This was 11th hour stuff, not reflected in exit polling.
There are six other states besides OH that "look funny" this time, and three or four others where there are odds and ends, that suggest the Bush Popular vote in them was lower than recorded. North Carolina, Virginia, and Louisiana were Southern states where pre-election polls showed Bush seldom above a Plurality even with a lead. Data from Oklahoma, California, Colorado, New Mexico, Nevada, Iowa, South Carolina and Texas also suggests Bush's vote numbers there may be partly bogus.
Even if Kerry didn't lead in the Popular vote, there is powerful evidence that he won in the Electoral College. And it's not just based on Ohio. Look at Florida, for gosh sakes, this time, as last time. Look at Nevada, a Bush Plurality state last time, where a judge threw out something, and only money prevents a third-party recount. New Mexico, which Gore carried last time. Iowa, with 4000 dead voters, with a Bush lead of 10,059 and all kinds of glitches.
So, this graphing of this, is another interesting angle on this.
If, say, history comes to show Bush had a lead of 760,000 to 1.9 million votes, and Kerry got over 300 Electoral College votes, I wouldn't be surprised. But would it ever make the national media.

[Rhapsody in Blue]

22. Fabulous graphs! Question -

Is there any way to identify the clock times along the bottom of the graph, to correspond with the percent of totals reported? I don't know if that info is available, but it would help to put the data into context with other events on that night.

Nice work!

:yourock:

[jbond56]

23. already working on it

In the process of collecting all the updates. It was updated about every 30 sec. I'm also recording what polls have closed. The only problem is it is a slow process.

[seaclyr]

24. clock numbers

aren't yet on the graphs, but they are listed in the data spreadsheet from which the graphs were constructed. I'll be happy to e-mail them out, just let me know.