Larger Audits Required to Confirm 2006 US House Races

[Wilms]

Larger Audits Required to Confirm 2006 US House Races

By Howard Stanislevic, VoteTrustUSA E-Voter Education Project

December 20, 2006

The verdict is in. This year's 401 contested US House seats resulted in closer electoral outcomes than the 400 seats that were up for grabs in the 2004 election. This means larger audits would have been necessary this year than in 2004 to actually confirm the results independently of electronic vote counting software.

For the following calculations, we assumed a 99% probability of miscount detection (confidence level), 400 precincts per Congressional District on average and the possibility of up to a 20% undetected vote shift per precinct.

Based on these assumptions, in 2004, 57 outcomes (14.3%) of the 400 contests required more than a 2% random audit of precinct totals to be confirmed. This year, using the same assumptions, 99 races (24.7% of the 401 contests) would require more than a 2% audit to confirm their outcomes.

While the above assumptions are not necessarily etched in stone (and do not take precinct size variations into account), it's clear that a small fixed percentage audit is even a worse idea now than it was two years ago.

snip

http://www.votetrustusa.org/index.php?option=com_content&task=view&id=2135&Itemid=26

[Contrite]

1. Exactly so and this is why O'Dell and Simon say hand counts only.

They said even a 10% audit of ALL precincts is inadequate except as an exit strategy from electronic vote counting.

[Bill Bored]

2. It is inadequate and it will NEVER tell you if ANY of your machines

are counting correctly or not.

I'd rather look at whole machines and whole scanners with errors as small as 1 vote then try to create the perfect "exit poll" using real ballots. But to each their own.

[Bill Bored]

9. Woops, should have said "THAN" -- not "THEN try to create the perfect 'exit poll...'"

It makes a BIG difference in this case! :)

[OnTheOtherHand]

3. I think he should have said

not "it's clear that a small fixed percentage audit is even a worse idea now than it was two years ago"

but "it's even more clear now than it was two years ago that a small fixed percentage audit is a bad idea."

(Of course, "bad" depends on the alternatives. I will say again that even a 2% audit (edit to add: i.e., a properly conducted 2% audit of reliable paper records) would place an upper bound on how many races could be hacked without a large risk that at least one of the hacks would be detected. That would be a huge improvement. But I don't think it's good enough.)

[IndyOp]

5. Have you seen Howard's online calculator?

It is quite nice and he has links to his papers at the bottom so people can understand the information they need to input and why.

http://mysite.verizon.net/evoter/AuditCalc.htm

[OnTheOtherHand]

7. yup, good stuff

I've actually talked with Howard a lot about his work -- you'll see me in the acknowledgements in his paper. Sometimes we have differences of emphasis, but there is very little that we really disagree about.

[Bill Bored]

11. Me too. He was probably just trying to appeal to the masses! nt

[truckin]

4. Good article. I'll be reporting soon on the results of audits that

took place in CT.

One of the big issues is what level of errors should trigger expanded hand counts. Also, what reasons are valid to accept audit differences and which reasons are not. It is critical to have strong audit standards but it is just as important to have specific guidelines that tell election officials when expanded counts are needed. It will be difficult to develop legislation that addresses this issue. Any ideas anyone?

[IndyOp]

6. Will you please PM me when you report the results of CT audits...

I am looking for examples of how the process can work so I can convince Indiana that IT CAN BE DONE!

As for setting "trigger points" for when hand counts should be required, you might want to check out Howard's online audit calculator - it asks for "max. within-precinct miscount as a % of total precinct vote" so he must discuss this in one of the papers he links to at the bottom of the page.

http://mysite.verizon.net/evoter/AuditCalc.htm

[truckin]

8. No Problem

[philb]

18. Why are only places with no indications of problems doing audits & places with indications of lots

of irregularities, not

[Bill Bored]

10. I think Triggers may be an "open research question."

But suppose there were two basic triggers:

Trigger

An error large enough to reverse the outcome all by itself, which would immediately trigger a full recount. Say Margin/2*Audit% in votes.

And Trigger Jr.

An error rate just above the expected (or acceptable) error rate of the voting system. It's used to trigger a larger audit but not a full recount until it reaches the size of the big Trigger.

[OnTheOtherHand]

12. excellent, but Trigger should be smaller

(depending on what you meant by margin/2 in votes.)

In most races, there will be a huge gap between the acceptable error rate and the rate that would flip the election. But if the election margin is 5%, and the observed error in the audit is 4.9%, that calls for a full recount!

I guess there are intermediate cases where the initial audit might leave doubt about the outcome, but a (say) 20% audit could pretty well clear things up.

[Bill Bored]

14. I said "margin/2*audit% in votes."

So let's say you audit 10% of a 100,000-vote race and the margin race was 10,000 votes. The recount trigger is: 10,000/2*.10 = 500 votes. If you have 500 errors or more in the 10,000-vote sample, it triggers a full hand count.

[OnTheOtherHand]

17. the part I was unsure about was "errors"

If candidate A loses 250 votes and candidate B picks up 250 votes, is that 500 errors? If so, this is probably fine, although I want to kick it a bit to see if anything could sneak through.

[Bill Bored]

19. Oh I see.

Well in practice, the audit would compare the machine tallies and hand tallies -- not individual ballots. So I'm saying that if the net difference in the margins of the audited votes were Margin/2*Audit%, it would trigger a full recount, which seems to be what you're saying.

The other question, which is related to Febble's point, is whether large errors that cancel each other out should trigger a full recount.

Suppose in half the audit we had a shift of 250 votes from Candidate A to B and in the other half, another 250 votes from B to A? What would that trigger? (I'm getting dizzy.) This might happen, for example, if ballot order rotation is not set up correctly.

[Febble]

13. At the risk of repeating myself...

even if the errors are within the error rate of the voting system, there needs to be a test as to whether significantly more of the go one way than the other. If so, Trigger Should Be Told.

[Bill Bored]

15. Details, details. nt

[Melissa G]

16. K&R for wilms!

:hi: