Democratic Underground Latest Greatest Lobby Journals Search Options Help Login
Google

Statisticians and elections specialists: If you had some really nasty data

Printer-friendly format Printer-friendly format
Printer-friendly format Email this thread to a friend
Printer-friendly format Bookmark this thread
This topic is archived.
Home » Discuss » Archives » General Discussion (Through 2005) Donate to DU
 
BevHarris Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 11:08 AM
Original message
Statisticians and elections specialists: If you had some really nasty data
come into your hands, data relating to the most bizarre disconnect between poll data and voting machines yet, data that would, if proven, force America to rethink everything about how states supposedly voted in 2000 --

Of course, you'd need the data vetted by academics. But are statisticians really qualified to interpret poll data discrepancies, or might you need someone with great expertise in polling and/or political science?

What credentials would you seek in experts to cast rock-solid opinions on this data, opinions that could stand up to scrutiny by lawyers for the mainstream press?

This one has the potential to do something big -- IF, that is, anything big can be made of poll vs. voting machine discrepancies at all.

What qualifications should I look for? What are the issues? How would you debunk a study that showed statistically impossible results between poll and voting machines, in a very surprising (and critical) location for the 2000 election?

Bev Harris
Printer Friendly | Permalink |  | Top
unblock Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 11:15 AM
Response to Original message
1. i'd look for an oscar winner. or maybe a grammy winner.
seriously. as a director of analytics for a securitization firm, i probably have the necessary math/stats credentials myself. but who would care what i think?

but sign up david duchovney or britney spears to carry the torch, well, now you're on to something.

not meaning to sound flip, but this is politics, after all....
Printer Friendly | Permalink |  | Top
 
Name removed Donating Member (0 posts) Send PM | Profile | Ignore Sat Jun-12-04 08:28 PM
Response to Reply #1
30. Deleted message
Message removed by moderator. Click here to review the message board rules.
 
LTR Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 09:05 PM
Response to Reply #30
33. Well, if you don't like this message board...
...find another.

You don't HAVE to post here, and I doubt you'll change any minds.

Congratulations! You just wasted time!

:eyes:
Printer Friendly | Permalink |  | Top
 
nolabels Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 09:26 PM
Response to Reply #33
35. Does seem like a lot time waste, doesn't it ?
If needed to waste my time on trying find people to hate I wouldn't be coming here either. I did occur to me once that more sane people would get to know others before they started slinging invectives. Be that as it may.

When you want people to agree with you, do try to intimidate them or do go to their wavelength to try to understand what they are thinking and why?

This might possibly be where you have gone wrong but maybe not,
............. besides what do I know anyway, everybody ignores me and I do prefer it that way :evilgrin:
Printer Friendly | Permalink |  | Top
 
LTR Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 09:29 PM
Response to Reply #35
36. Yeah, it's pretty bed when...
...you drive a Viper and you spend Saturday nights spamming message boards.

:eyes:
Printer Friendly | Permalink |  | Top
 
nolabels Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 09:42 PM
Response to Reply #36
38. I will send him a post card as he passes into the distance
Printer Friendly | Permalink |  | Top
 
LittleApple81 Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 10:06 PM
Response to Reply #33
41. This guy... In my country they say: If you say goodbye so many times, you
really don't want to leave.
I have seen his posts on other threads already. At least he put two of them together into a longer one!
Printer Friendly | Permalink |  | Top
 
maryallen Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 10:01 PM
Response to Reply #30
40. Spam alert!
Edited on Sat Jun-12-04 10:03 PM by maryallen
Isn't it amazing that on this very important thread that a so-called "disgruntled poster" spams the same message he/she spouted in previous threads?!

Enough said --

Back on message:
It couldn't possibly be Texas ("very surprising and critical state"), could it, Bev?
Printer Friendly | Permalink |  | Top
 
unblock Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 11:16 AM
Response to Original message
2. oh, forgot something very important:
:yourock:
Printer Friendly | Permalink |  | Top
 
Eric J in MN Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 11:17 AM
Response to Original message
3. If you're talking about polls, you would need someone with
If you're talking about polls, you would need someone with expertise in polls, which many statisticians have.
Printer Friendly | Permalink |  | Top
 
Snow Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 11:18 AM
Response to Original message
4. Okay, I'm an epidemiologist but have a master's in biostatistics
and teach introductory grad-level courses. Statistical techniques allow you to state whether the finding you've observed is a result of chance, with a probability assigned to that statement. Anytime you see an association, the name of the game is ruling out alternative explanations, and the explanation of "chance occurrence" is one of the first things you want to rule out. Therefore, it's unfortunate that a bit of statistical jargon is the term "significant". That doesn't mean "important"; it simply means the degree to which you can exclude chance as an explanation. A very low probability of chance as an explanation doesn't mean you'v explained the finding yet. So, yes, you need a statistician, and yes, you do need poli-sci, polling, or maybe even better, a geographer, since geographers study this kind of thing all the time.

When you go looking for a statistician, I'm betting you'll need someone who knows how to deal with messy data - ie, the data may be sparse in some spots, it may not follow a normal distribution, there may be dependence among different variables. Who ya gonna call? Well, someone who knows and wouldn't mind using bootstrap, resampling, and Monte Carlo analytic techniques. Maybe even (and here half the statisticians in the world will shudder in horror) someone who's a bayesian. Good luck!
Printer Friendly | Permalink |  | Top
 
unblock Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 11:24 AM
Response to Reply #4
7. this is why i think i have the necessary background
basically, my job is to price a company's invoices. so i have to look at their billing an collections history. often the data is coming from different systems, handled or coded or stored in different ways from time to time or from division to division.

what a mess, but hey, that's why they pay me the big bucks.

bev, i would LOVE to help out any way i can. i'm sure i can help with the math/stats/analysis. my only concern is, as i said before, who would listen to me?
Printer Friendly | Permalink |  | Top
 
BevHarris Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 04:45 PM
Response to Reply #4
17. Thanks. I learned something from your post
right before my brain popped out of my head.

Bev
Printer Friendly | Permalink |  | Top
 
Snow Donating Member (1000+ posts) Send PM | Profile | Ignore Mon Jun-14-04 10:49 AM
Response to Reply #17
50. You're quite welcome....
look for a statistician who knows what my post meant - if she/he gets the joke about bayesians, then that's a good candidate.
Printer Friendly | Permalink |  | Top
 
Lisa Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 05:04 PM
Response to Reply #4
20. geographers are handy if you need spatial stats ...
They've got ways to take things like spatial autocorrelation into account. (I have a PhD in physical geography but I don't consider myself a stats expert by any means, even after teaching methods courses!) If you've got someone who's a GIS/geomatics specialist, maybe, but not everybody with that background will have experience in electoral data. Census, maybe (if they're from the social science side rather than the physical). But if you can find someone who has a solid background in political geography looking specifically at elections, and who has plenty of hands-on in stats and/or Geographical Information Systems ....

Peter Gould would be perfect (particularly with his strong interest in social justice issues) but unfortunately he passed away a few years ago. He was at Penn State, I think.
Printer Friendly | Permalink |  | Top
 
Snow Donating Member (1000+ posts) Send PM | Profile | Ignore Mon Jun-14-04 10:50 AM
Response to Reply #20
51. Right - but what I had in mind was more
that Bev needs a couple or more people who can collaborate on this effort - and I think a geographer and a statistician should be in the mix.
Printer Friendly | Permalink |  | Top
 
displacedtexan Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 11:21 AM
Response to Original message
5. Doctoral level demographers with hands-on experience
You also need someone who is fluent in several programming languages and who can invert matrices at the drop of a hat.

PM me if you want to know more. I can hook you up with someone who evaluates Justice Dept. programs.
Printer Friendly | Permalink |  | Top
 
TruthIsAll Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 11:23 AM
Response to Original message
6. Bev, here is my 2002 Senate election fraud analysis..
Edited on Sat Jun-12-04 11:49 AM by TruthIsAll
It's simple probability and statistics 101.

You can do it using probability analysis to present a heavily circumstantial case for fraud. Lawyers use mathematical studies all the time to determine the probability of a series of unlikey events.

My credentials:
BA Mathematics
MS Applied Mathematics
MS Operations Research

This is a simple method to determine the probability that 4 out of 10 critical states, in which the Dems were far ahead (beyond the margin of polling error), would all go for the Repub.

One can apply the Cum. Normal Distribution or Poission function (or both to confirm) to determine the probability of rare events occuring by chance only.. There are other models which may be applicable. I found these two perfect and easy to use. All you need is Excel; the functions are built-in.

I have also used this method to calculate the probability that
1- at least 15 JFK witnesses would meet unnatural deaths in the year following the assasination.
2- at least 16 world-class microbiologists would meet unnatural deaths in a 4 month period following 9/11.
3- The probability that at least a certain number of people would suffer from mad cows disease in a specific geographic area in a given year.

In each case, the probabilities were close to zero.


In the 2002 election, the odds are less than 1 out of 43,000 that the results were due to chance.

In other words, highly circumstantial evidence of fraud, using classical probability theory.

Basically, we know the probability of a given election falling outside the polling margin of error, say +/-3%. That is 1 out of 20.

In other words, if candidate X has 53% in the final polling, and Y 47%, the chances that Y would win in a fair upset are 1 out of 20.
The MOE means that X would get between 50% and 56% of the vote 95% of the time or, similarly, Y would get between 44% and 50% 95% of the time.

You can go from there.
http://www.democraticunderground.com/discuss/duboard.ph...
Printer Friendly | Permalink |  | Top
 
ThoughtCriminal Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 01:41 PM
Response to Reply #6
14. A question though
:evilfrown: Here
Disclaimer: Not a statistician, not even on TV, did not stay at Holiday Inn Express..

If 1/20 elections fall outside the polling margin of error and there are ~465(?) contested elections then it would not be surprising if 20+ fell outside the margin.

But there is another way of looking at it. Do all of the anomalies favor a particular political party? Unless there is some demographic or similar explanation (and there may be some arguments for that) we would expect that elections that fall outside the polling margin should not favor one party over another. In 20 cases, we would expect an average of 10 to favor Republicans, 10 to favor Democrats.

If there are 20 anomalies and all 20 favored one party, I think the probability would be 1 / 2^20 (see disclaimer above).

Now - How many of these anomalies favored Republicans?


Printer Friendly | Permalink |  | Top
 
TruthIsAll Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 08:22 PM
Response to Reply #14
28. Go to the link. 4 out of 8 hard fought senate elections all went Repuke
Edited on Sat Jun-12-04 08:31 PM by TruthIsAll
even where the Dems held big leads in excess of the MOE.

Even where the final polling was close in Texas, with 1% separating the candidates, the Repuke won by a whopping 12%, much higher than the margin of error. The odds of this were significantly lower than 1 out of 20, which is based on exceeding the 6% MOE range margin (the difference between +3% and -3%).

The Repukes not only beat the MOE, they beat the MOE big-time.

I guess they wanted to make damn sure the elections were stolen. y having the elections appear to be landslides, they probably figued that there would be no reason to question the results.

Printer Friendly | Permalink |  | Top
 
ThoughtCriminal Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 09:36 PM
Response to Reply #28
37. OK I think I get it - more questions
And possible areas for research perhaps.

There 34 Senate seats up for election. Were only 8 "Hotly contested" - what is the criteria?

How many of the "Not Hot" races fell outside the MOE?

Historically, are "Hot" races more or less likely to fall outside polling MOE? If there is a significant difference can it be explained by voter behavior?

Historically, are MOE anomalies more likely to take place where the prevailing party has greater control over the election process?

Were there MOE anomalies favoring the Republicans in House races in the same states?

Have the number of MOE anomalies changed over time?


Printer Friendly | Permalink |  | Top
 
TruthIsAll Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 10:22 PM
Response to Reply #37
42. OK, I will be brief.
Edited on Sat Jun-12-04 10:26 PM by TruthIsAll
There 34 Senate seats up for election. Were only 8 "Hotly contested" - what is the criteria?
-Hotly contested means those elections which both parties were concentrating on for senate control; the others were more or less a foregone conclusion. For example, Kerry in Mass. See the Scoop link

How many of the "Not Hot" races fell outside the MOE?
- Don't know, because they were of no consequence. If Kerry polled 80%, and ended up with 75%, no one would notice. Or care.

Historically, are "Hot" races more or less likely to fall outside polling MOE? If there is a significant difference can it be explained by voter behavior?
-No. These were anomalies. "Hot" races are the competitive ones. These are the races which are closely watched. And the polling MOE is within range of the result 95% of the time. To have 4 of eight fall way outside the MOE is statistically extremely remote. If it were one of eight, OK. But four?. No way.

Historically, are MOE anomalies more likely to take place where the prevailing party has greater control over the election process?
- I do not know. But I would assume so, knowing what we now know.

Were there MOE anomalies favoring the Republicans in House races in the same states?
- Check the scoop link.

Have the number of MOE anomalies changed over time?
- 2002 was special. Zogby said he never saw anything like it. He is the most accurate, yet his final polls were so far off the mark, he had no explanation for it. At least, that's what he said. I believe he suspects foul play, but of course he would never say so.
Printer Friendly | Permalink |  | Top
 
ISUGRADIA Donating Member (1000+ posts) Send PM | Profile | Ignore Mon Jun-14-04 12:21 AM
Response to Reply #6
49. You miss a key set of facts: Polls are only a snapshot of the moment
and polls are only as good as the sample universe a pollster has used in his or her polls.


I know in the past you have cited polls that were a week, or several days out to show that election day results are suspect. But one of the first things I learned in stats was that the poll is as only a snapshot of that moment. So a poll even a day before the election is not necessarily a good predictor of the results. Trends can be recognized that show momentum towards a candidate.

The only poll that counts is the one on election day. Exit polls turnout to be fairly accurate predictors of results because they are sampling voters, the actual universe a pollster needs to draw from. Up to that time a pollster is relying on self-identified probable or likely voters. Which gets to my second point...


Pollsters' polls are only as accurate as the sample they have developed. If Democrats are overrepresented in a poll compared to the results on election day, for example, the results will be skewed. Have you ever examined actual turnout rates for parties v what the sample universe for polls you have used were? This is what pollsters site in most cases for errors they make.

For example, if I construct my polls the weekend before an election expecting the party breakdown to be 35% D, 35% R, and 30% No Party and exit polls show the actual breakdown is 30% D, 38% R, and 32% No party, of course results may not reflect pre-election day poll results. The sampling did not reflect the actual universe of voters.

This can be used for other variables such as race, larger than expected African-American turnout in 1986 caused a large number of too close to call races to go D that year in the South.


I don't think you can determine the probabilty of Senate races going for one party or another in 2002 or any other year from any polls that were not taken ON ELECTION DAY. Older polls are irrelavent statistically speaking to determine probability because they only are making an educated guess as to the universe to sample from. So I think your basic premise is flawed.


Printer Friendly | Permalink |  | Top
 
alittlelark Donating Member (1000+ posts) Send PM | Profile | Ignore Mon Jun-14-04 11:44 AM
Response to Reply #49
53. exit polling is highly accurate
ISUGRADIA wrote,

"I don't think you can determine the probabilty of Senate races going for one party or another in 2002 or any other year from any polls that were not taken ON ELECTION DAY. Older polls are irrelavent statistically speaking to determine probability because they only are making an educated guess as to the universe to sample from. So I think your basic premise is flawed."

Isn't it odd that exit polling was cancelled for the 2002 election?

It had always been highly accurate - EXCEPT in the 2000 election.

odd, huh.
Printer Friendly | Permalink |  | Top
 
ISUGRADIA Donating Member (1000+ posts) Send PM | Profile | Ignore Mon Jun-14-04 01:01 PM
Response to Reply #53
57. Clarification
2002 exit polling was not cancelled but was not used by networks on election day because of a breakdown of the VNS system. That data does still exit, I believe, for 2002 exit polls. I'd have to do checking to see if it is available on the Internet.


There were flaws in VNS modelling in 2000 and interpretation of data in the key state of Florida, but I do not remember other key Senate races or states being called incorrectly based on exit poll data.


Exit polling is not perfect either, it's just a sampling of the larger universe. I do think examination of exit polls would show a very large number of races within the margain of error and, for the most part, fairly accurate if good sampling is employed.


I do think it is important for examination of possible patterns of fraud, tampering for local experts familiar with county, city voting patterns to be consulted and used.

Printer Friendly | Permalink |  | Top
 
KoKo Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 11:32 AM
Response to Original message
8. Good luck on this Bev, if it's going where I think it's going, the truth
will be known. Wish I could help you, but imagine someone here will give you some good leads.
Printer Friendly | Permalink |  | Top
 
BeFree Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 11:34 AM
Response to Original message
9. Hey Mike
Edited on Sat Jun-12-04 11:35 AM by BeFree
Whatta ya think? Know any FBI people who could look at Bev's data? Do us a favor, big guy. You're job is to look at ALL the facts, right? Have a look at this and tell us what ya think.

Printer Friendly | Permalink |  | Top
 
ContinentalOp Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 12:07 PM
Response to Original message
10. I'm not an expert,
But if there's not one big impressive name you could rely on, perhaps you could run the numbers by several different people so you could say that the same conclusions were reached independently from a number of experts. Just a thought.
Printer Friendly | Permalink |  | Top
 
unblock Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 12:56 PM
Response to Original message
11. bev, have you tried famous election prognosticators
like larry sabato?
Printer Friendly | Permalink |  | Top
 
Eloriel Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 05:10 PM
Response to Reply #11
22. Instead of that
Edited on Sat Jun-12-04 05:10 PM by Eloriel
how 'bout some of the folks who were involved in the technical aspects of Voter News Service?

(Larry Sabato is a rightwinger -- not someone I'd want any advice from, about anything, frankly. And yes, that's extremely biased. That's the way I am. ;-) )

Edit: I just remembered who it is who knows someone who was with VNS -- Mike Malloy's wife, Kathy.
Printer Friendly | Permalink |  | Top
 
BevHarris Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 05:16 PM
Response to Reply #22
24. Really? Those are great suggestions.
I think, specifically, the problem is this: at what point does a polling pattern that mismatches a set of voting machine results become

a. Statistically significant
b. Statistically bizarre
c. Statistically unexplainable
d. Statistically impossible

And how do you take that kind of discrepancy and "prove something" with it? Because, of course, even if as Dr. Henry Lee so famously said, "something is wrong" -- it could be wrong with the voting machines, or it could be wrong with the polls themselves.

Very good insights, thanks Eloriel.

Bev
Printer Friendly | Permalink |  | Top
 
zelda7743 Donating Member (256 posts) Send PM | Profile | Ignore Sat Jun-12-04 01:03 PM
Response to Original message
12. Give it to Stephen Hawking
Everyone believes him. :)
Printer Friendly | Permalink |  | Top
 
ConsAreLiars Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 01:29 PM
Response to Original message
13. Three standards for "expertise"
One is scientific endorsement, like publication in a peer-reviewed journal, or better a review commission set up by http://www.aaas.org /.

Another is street creds - a known credible advocate smart enough to "get it," someone with celebrity points, to go around and explain it. Like Jimmy Carter or Ben Cohen or Paul Newman or Lou Dobbs even.

The third is mass opinion ("everybody says that ..."), just getting the news well known via media and political campaigns. True Majority, MoveOn, Oprah Winfrey, Howard Stern ....

The first hurdle is today's challenge. Allan Lichtman (a "historian" not a statistician) did the analysis of black voting disenfranchisement in Florida http://www.usccr.gov/pubs/vote2000/report/main.htm and http://permanent.access.gpo.gov/lps13588/lps13588/ch1.h... that revealed that the odds of getting votes invalidated was correlated with the proportion of black voters in the districts. The report was drowned out by the Reich-wing noise machine in part because it could be attacked as "opinion." But Lichtman probably has contacts among statisticians who might form a team, and putting it out first as an academic study rather than a political commentary might give it better traction.
Printer Friendly | Permalink |  | Top
 
RedEagle Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 03:28 PM
Response to Reply #13
15. Florida...
...which reminds me of Greg Palast, might see if he is hooked up with anybody.

And I'd keep options for getting help from outside the US, too.
Printer Friendly | Permalink |  | Top
 
ConsAreLiars Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 03:35 PM
Response to Reply #15
16. He's on the board of votewatch
http://www.votewatch.us/advisors.php

Along with Lichtman and several others who might(?) be interested in taking up this analysis and giving it a push.
Printer Friendly | Permalink |  | Top
 
BevHarris Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 05:05 PM
Response to Reply #16
21. Yes, VoteWatch is hooked up with statistical analysis stuff
I started to call Steven Hertzberg, who heads that group. What blocks that is lack of agreement on how to collaborate, among voting groups.

One of the high priorities over the next few weeks has to be getting past turf protecting behavior so that we can work cooperatively on projects like this.

That's been a problem in the past, but we are making headway. This is something that I've been proposing to many of the groups already -- and am starting to get sign-on for: We need to find much better ways to play nicely with our friends, rather than trying to snatch the ball and take it home to keep it for ourselves.

Cooperative work must particularly come into play with election monitoring activities -- at least seven groups that I know of are planning to do it, each focusing on a different thing, actually, but to date none of them have done much to coordinate with the others. (That was MY report. No, it was MINE FIRST. Well WE got one too! I told my donors that YOU SUCK. Those people don't know what they're doing. Nyah nyah.) I had a one-hour phone conversation yesterday with one of the major players, discussing exactly this. I think we made very good headway.

The biggest obstacle to mobilizing our resources more effectively, in fact -- whether it is collaborating on statistical analysis, like this, or on election monitoring procedures and hotlines, is that every group, for understandable reasons, wants to keep their sources privately held and wants to be the one to break the story. But those obstacles can be solved (if we have the will to do so) by alliance agreements which lay out the rules for sharing.

Safeguarding the election is just too important to erode into a bunch of piddly-ass turf wars.

Printer Friendly | Permalink |  | Top
 
Rebellious Republican Donating Member (1000+ posts) Send PM | Profile | Ignore Mon Jun-14-04 07:53 PM
Response to Reply #21
59. Good luck Bev, I have been preaching co-operation for two years
here at DU. It is the only way that we are going to beat the Republican machine. I know, I used to be one! If only long enough to bring down the neocon branch of the republican party, thats all then we can go back to open healthy debate that we democrats cheerish so much. But many just do not get it.




Printer Friendly | Permalink |  | Top
 
Rebellious Republican Donating Member (1000+ posts) Send PM | Profile | Ignore Mon Jun-14-04 07:55 PM
Response to Reply #59
60. And, when Kerry becomes President, I hope he puts you in .....
charge of the Federal Elections Commission.

:yourock:




Printer Friendly | Permalink |  | Top
 
BevHarris Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 04:49 PM
Response to Reply #13
18. Thanks -- good suggestions.
Very good feedback, giving me some good directions to explore.
Printer Friendly | Permalink |  | Top
 
Cheswick2.0 Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 04:54 PM
Response to Original message
19. Bev bless you for all that you do
I would get a statics expert and a political expert to work together.

I am going to guess that the location you are talking about is NH, MO or TN and I know you can not tell me.

Good luck and keep us posted.
Printer Friendly | Permalink |  | Top
 
Jack Rabbit Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 05:15 PM
Response to Original message
23. As a student of philosophy and logic and with a BA in polisci
Edited on Sat Jun-12-04 05:21 PM by Jack Rabbit
Of course, I'm absolutely convinced that Katherine Harris stole the election for Bush in Florida, but that has nothing to do electronic voting machines. That had to do with purging voter rolls.

Questions I would ask:

What is the size of the sample of polling versus results discrepancies?
How reliable were the polls that where "shown" to be "wrong"? What are the historical trends of these polls against actual polling results?
In cases where opinion surveys prior to the election differed greatly from the actual election results, in what percentage were voting machines used?
Is there a correlation between the size of the discrepancy and the use of electronic voting machines?

Those questions are designed to root out the probability of bad polling data prior to the election or polls that are historically inaccurate. Such polls are meaningless, with or without BBV. The questions also assume that you are examining voting data in areas that used conventional voting methods as well as those that used BBV. If you don't have that, your data won't prove that BBV is no different from or any better or any worse than conventional voting.

If the polls were more wrong than usual in places where electronic voting was employed and the correlation coefficients show that there is a wider discrepancy in pre-election surveys vs. actual election results in places where electronic voting was used as opposed to more conventional voting methods, then you ar justified to think that you are smelling a rat. The strength of the odor is directly proportional to the correlation of wider discrepancy to electronic voting.

If I were looking for a good statistician or mathematician I would go to a university and ask for help from the chair of the mathematics department.
Printer Friendly | Permalink |  | Top
 
BevHarris Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 05:18 PM
Response to Reply #23
25. So clearly explained that I almost feel I could do it myself!
And then, of course, I came to my senses. Thanks a million.

Bev
Printer Friendly | Permalink |  | Top
 
Jack Rabbit Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 05:50 PM
Response to Reply #25
26. You're welcome
I am happy ot assist in your cause.
Printer Friendly | Permalink |  | Top
 
KittyWampus Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 08:11 PM
Response to Reply #23
27. Trying To Understand All This
:kick:
Printer Friendly | Permalink |  | Top
 
Jack Rabbit Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 08:48 PM
Response to Reply #27
31. It works like this
Bev is attempting to show that there is a discrepancy between pre-election polling data and the actual results and relate this phenomena to the use of electronic voting machines. To do this she will need to show, first of all, that in areas where BBV was employed there is a wider discrepancy between public opinion surveys and election results than in areas with other methods of voting and that this wider discrepancy is statistically significant; then she will have to show that isn't just bad sampling by the poll takers or that it could be attributed to anything else but the use of BBV.

That might not prove anything beyond a shadow of a doubt, but it would be evidence that there is something amiss.

In then case of the Florida 2000 election, this might be difficult. Off the top of my head, I might attribute some discrepancy there to the fact that tens of thousands of voters who thought they were registered to vote found out when they went to the polls that they were not. That wouldn't have anything to do with BBV; that would be Katherine Harris illegally purging the voter rolls.

I haven't seen Bev's data and have no idea what it says or what she thinks it might say. However, if she wants to make a case, that is what she needs to do.
Printer Friendly | Permalink |  | Top
 
RedEagle Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 11:54 PM
Response to Reply #31
43. It's still computer counted
BBV is just part of the issue.

Even when we use paper, it's not audited and the reliance is totally on the computer count.

All voting systems must produce a voter verified paper ballot- then the ballots have to be audited by hand in random audits.

So, working on data with all voting systems is relevant.

For example, there's the 16,022 negative votes for Gore in Florida. That came from an optical scan system. How many more times did that happen and not get caught?

Punch cards were the distraction. They were desperate not to have the optical scan counted.

I wonder if Zogby could refer you to someone?
Printer Friendly | Permalink |  | Top
 
ShaneGR Donating Member (1000+ posts) Send PM | Profile | Ignore Mon Jun-14-04 11:58 AM
Response to Reply #23
54. Everything he said until the last line....
Edited on Mon Jun-14-04 11:59 AM by sgr2
You need a professor who teaches a statistics related Grad Level Political Science class. The class here is called "Scope & Theory." These classes are normally database application classes revolving around probability in numbers. You need to "prove" that polling data prior to an election was "correct" and the election results were "incorrect." To do that you need to prove that the polling firms responsible for the "correct" polls have a spotless record going back decades. You need to prove that the "incorrect" election results are indeed "incorrect." To do that, you need to be able to prove that nothing happened in the last few days to statisticaly swing the election "X" points in the other direction. For that you need the spotless polls, the election results, AND the exit polling data. The exit polling must come close to matching the polls taken prior to the election.

That's a tall order.
Printer Friendly | Permalink |  | Top
 
bearfan454 Donating Member (1000+ posts) Send PM | Profile | Ignore Mon Jun-14-04 12:11 PM
Response to Reply #23
55. What about residuals from the data ?
They mean a lot.
Printer Friendly | Permalink |  | Top
 
Cocoa Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 08:27 PM
Response to Original message
29. get someone to give you the conclusions you want
Printer Friendly | Permalink |  | Top
 
AP Donating Member (1000+ posts) Send PM | Profile | Ignore Sun Jun-13-04 08:39 AM
Response to Reply #29
45. Why are you still so skeptical?
Printer Friendly | Permalink |  | Top
 
Trajan Donating Member (1000+ posts) Send PM | Profile | Ignore Mon Jun-14-04 11:29 AM
Response to Reply #29
52. Data is data ...
Data doesnt have personality ...

IF we are wrong about BBV, and it turns out that NO fraud is occuring: then Im sure each and every one of the BBV crew would be glad to admit that ....

Statistical analysis is one means of obtaining such evidence: ANYONE is free to do such an analysis ... interpreting the results can be subjective, but the statistical community itself could verify such an analysis and come to a reasoned, consensus conclusion to the validity of such analysis ...

That is as good as it gets in this world ... whether someone 'wants' to get specific conclusions or not ...
Printer Friendly | Permalink |  | Top
 
Blue Wally Donating Member (974 posts) Send PM | Profile | Ignore Sat Jun-12-04 08:50 PM
Response to Original message
32. Sometimes People Say Different Things To Pollsters
One primary season back in the sixties or seventies, George Wallace, the Alabama "populist" polled an amazingly high number of votes in the Michigan Democratic Primary. No one told the pollsters they were voting for Wallace. No one told their friends they were voting for Wallace. On election day, they all went into the voting booth and pulled the lever for George Wallace.
Printer Friendly | Permalink |  | Top
 
SharonAnn Donating Member (1000+ posts) Send PM | Profile | Ignore Sun Jun-13-04 06:37 PM
Response to Reply #32
48. The Michigan Dutch Reformed are very racist, but they don't
admit it. So, it makes perfect sense to me.

The Michigan Dutch Reformed side of my family has to be heard to be believed.
Printer Friendly | Permalink |  | Top
 
JanMichael Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 09:08 PM
Response to Original message
34. If I couldn't do it I know people who could.
Extrapolation of elections data to other data sources is fairly simple for those with the right software.
Printer Friendly | Permalink |  | Top
 
bluestateguy Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 09:52 PM
Original message
Try the American Political Science Assn. Convention this Sept.
Sept. 2-5 in Chicago. Although the program is already set. Usually the big east coast papers send a reporter to cover it.

www.apsanet.org

Printer Friendly | Permalink |  | Top
 
dsc Donating Member (1000+ posts) Send PM | Profile | Ignore Sat Jun-12-04 09:52 PM
Response to Original message
39. Just about any statistician is going to be familiar with
polling and setting up questions. I think your best bet would be actuaries or statisticians. An interesting choice might be biologists or medical researchers. Both of those would have experience in testing null hypothesis and in setting up multi variable experiments.

As to your second question. The best case would be an ability to compare an election with the machines and one without in which the issues, the candidates, and the voters were substantially similar. Ohio could provide an ideal labratory for this but it would be too late to help. Here it is county by county and only some of our counties have the machines. Thus if you found two counties that were similar enough you would have the other variables controlled for. Barring that keeping the voters the same, the type of election the same, but only changing the time would be a distant second.
Printer Friendly | Permalink |  | Top
 
T Roosevelt Donating Member (1000+ posts) Send PM | Profile | Ignore Sun Jun-13-04 07:54 AM
Response to Original message
44. Other data to consider
- voter turnout by registration - looking for skewed turnout numbers

- past voting trends by region (eg county) - anomalous crossover vote counts

Here is some material I found that might help on this - it's the first chapter of a document (the rest is not on the website). The contact might also help. It seems more strategy (at least from this chapter) but may lead to other solutions:

Mr Richard Klein
Senior Programme Manager, National Democratic Institute (NDI), Washington D.C.

Quick Count - Chapter 1
Printer Friendly | Permalink |  | Top
 
RedEagle Donating Member (1000+ posts) Send PM | Profile | Ignore Sun Jun-13-04 05:04 PM
Response to Reply #44
46. Kick
Printer Friendly | Permalink |  | Top
 
buycitgo Donating Member (1000+ posts) Send PM | Profile | Ignore Sun Jun-13-04 05:10 PM
Response to Original message
47. Obligatory "I love you, Bev" post
can't praise you often enough

how many authentic American heroes hand out here, after all?

I talk about BBV all the time to anybody I can keep listening long enough to harangue, and most of them don't have a clue

they're usually very interested, and PO'd when they hear the deal

very often dont'/can't believe this is possible

thanks, again, Bev

you ARE the one!
Printer Friendly | Permalink |  | Top
 
Lorien Donating Member (1000+ posts) Send PM | Profile | Ignore Mon Jun-14-04 12:27 PM
Response to Original message
56. Film Producers use Statistitians
To determine how they market their films. I think I still have Jeffery Katzenberg's number (Dreamworks SKG); he's a big Dem contributer and friend of Gore's. He might be able to help. If you would like his number, please PM me.

Wish I had more to offer.

:shrug:
Printer Friendly | Permalink |  | Top
 
loudsue Donating Member (1000+ posts) Send PM | Profile | Ignore Mon Jun-14-04 07:19 PM
Response to Original message
58. kicking again in case a newcomer has some ideas!
:kick: :kick: :kick: :kick: :kick:
Printer Friendly | Permalink |  | Top
 
DU AdBot (1000+ posts) Click to send private message to this author Click to view 
this author's profile Click to add 
this author to your buddy list Click to add 
this author to your Ignore list Tue Oct 21st 2014, 09:06 PM
Response to Original message
Advertisements [?]
 Top

Home » Discuss » Archives » General Discussion (Through 2005) Donate to DU

Powered by DCForum+ Version 1.1 Copyright 1997-2002 DCScripts.com
Software has been extensively modified by the DU administrators


Important Notices: By participating on this discussion board, visitors agree to abide by the rules outlined on our Rules page. Messages posted on the Democratic Underground Discussion Forums are the opinions of the individuals who post them, and do not necessarily represent the opinions of Democratic Underground, LLC.

Home  |  Discussion Forums  |  Journals |  Store  |  Donate

About DU  |  Contact Us  |  Privacy Policy

Got a message for Democratic Underground? Click here to send us a message.

© 2001 - 2011 Democratic Underground, LLC