Re: Measurement Error and the 2000 US Election

Topics: Democracy
11 Nov 2000

From: Ervan Darnell

Good to have some traffic.

At 01:20 PM 11/11/2000 , Clive Woodward wrote:
>I think that the presidential election was a tie given that the difference
>in the final tally, whatever it turns out to be, will be much less than the
>measurement error inherent in the official estimate of the true vote.

Agreed.

From a public-choice point-of-view (not one I agree with however), it doesn't matter who wins. The public has said that value of the two candidates is identical. The cost of the wrong decision in Florida is only equal to the grief suffered by the number of people who made the marginal difference.

>[ the hand count will change the totals again and only shows the necessary precision is unavailable ]

Maybe, but this is an interesting theory to be tested. If the hand count comes within 10 votes of the machine recount, I'd call that substantial confirmation. I'm worried that the opposite will be the outcome and that the error rate will go up with hand counts because ballots will be destroyed or invalidated with new smudges each time they are handled. There is a certain amount of leeway in interpreting "almost holes" that the machine reader probably averages out fairly over a large number of ballots, whereas human counters let the biases prejudice the reading of dubious ballots. Regardless, it remains to be tested empirically (or maybe it has been tested elsewhere, any references anyone?).

And, yes, of course, whatever the prevailing law is in Florida should apply, if it addresses such issues as the legality of calling for hand recounts.

>My modest proposal (for 2004) is to disband the electoral college in favor
>of the popular vote and have run off elections between the two highest
>polling candidates if no one gets 50% of the vote. This is a generally
>accepted standard in modern non-parliamentary systems. The probability of a
>run off election could make people feel freer to vote for third party
>candidates in the initial election, which could make the process more
>interesting and informative but unfortunately also makes Republicans and
>Democrats less likely to propose it. The run off election also has the
>effect of manufacturing an artificial "mandate" that is at least as
>credible as that created by the electoral college.

Two good points in support of the change. I'm undecided on this matter. I've slowly reversed my previous position that 3rd parties need a bigger voice in the electoral process. I think the current system ruthlessly seeks the middle, and the introduction of more parties won't change that. One has to think the free advertising that comes with a campaign would "wake people up" to some particular point of view in order to see much value in a larger 3rd party presence.

Better would be to have approval voting, i.e. you can vote for as many candidates as you want. Better still is to allow open bidding for candidates (as many people on this group have suggested and the best analysis of which I knows comes from David Friedman's "The Machinery of Freedom").

For my part, I find even the intelligent, well-compensated people I work with generally hostile to the idea of liberty. Putting the LP in the national debates wouldn't change that.

>However we go about it, we need to know the measurement error of the vote
>counting process and we need to minimize that error. Fortunately that is a
>discoverable characteristic of the process. Any vote close enough to fall
>within that margin is really an undecided election and should be recognized
>as such -- pretending otherwise is intellectually dishonest.

The observation is accurate, but I'm not sure I buy the conclusion. Arbitrarily deciding for a given candidate in a tied election is no worse than having an arbitrary cut-off point for declaring a tie. You have to believe that a run-off somehow increases the resolving power of the election process to discover people's true intent. Even if you believe that, it just begs the question of how small the delta before declaring a run-off. In cases like this where the probable outcome of run-off is known, declaring a tie is just as loaded as declaring a winner.

I think my previous reason applies to *this* run-off as well as interpreting the popular vote. Nader's presence took the far left from Gore. Gore compensated in both directions (at least theoretically) by offering some programs to regain those on the left without too alienating those in the middle, and vice-versa. Thus, Bush was politically assaulted for the moderate vote that he would not have otherwise been. His battle was then to capture the rightmost 48% of the remaining electorate after ignoring the leftmost 4%. To have another vote is to allow Nader and Gore to have effectively run together. Bush's strategy would have been different had that been the case before-hand.

That's not an argument against changing the system as you previously outlined, because then everyone would know the rules going in. It's just an argument against ordering a new vote now. If we change the system to allow run-offs, it should be under clear guidelines (like your first suggestion of not obtaining an absolute majority), and not under some nebulous definition of a tie.



-------------------------- eGroups Sponsor -------------------------~-~>
eGroups eLerts
It's Easy. It's Fun. Best of All, it's Free!
http://click.egroups.com/1/9698/4/_/220122/_/973989940/
---------------------------------------------------------------------_->



Home