Re: "reasonable probability"
Christian Bau wrote: > In article <90k3vl$[EMAIL PROTECTED]>, [EMAIL PROTECTED] > (Herman Rubin) wrote: > > > >AFAIK there is general agreement that unbiased humans are better at > > >identifying the difference between unpunched holes and imperfectly > > >punched holes than current counting machines -- which after all were > > >only designed to distinguish between unpunched and perfectly punched > > >holes. > > > > UNBIASED humans, yes. There is considerable evidence of bias. > > > > Some of this bias is inadvertent, the type of observer bias > > found in many experimental situations in other fields. This > > is especially the case if it is not merely a piece of hanging > > chad, but a dimple. It also occurs if there is a question > > of multiple voting for an office. > > Both the United Kingdom and Germany use the old fashioned > piece-of-paper-take-a-pencil-mark-your-candidate method, and the papers > are always handcounted. So does Canada. We had municipal elections a few weeks ago; the ballots were paper, with boxes for making an X with a standard, supplied pencil.. These were then fed into a machine for automatic counting. Since such optical recognition technology is several decades old, why not just use it? In our recent Federal election there was one office to vote for in every polling station, by making an X in a box. The country-wide counting was finished by early morning. Counting is done by people, with "scrutineers" from each of the political parties present at the count. > > > I don't think there is ever any question about "voter intention" unless a > voter deliberately chooses to make his ballot paper undecidable. Just > because a system is oldfashioned doesn't mean it can't be better anyway. -- R. G. Vickson Department of Management Sciences University of Waterloo Waterloo, Ontario, CANADA = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: "reasonable probability"
In article <[EMAIL PROTECTED]>, Christian Bau <[EMAIL PROTECTED]> wrote: >In article <90k3vl$[EMAIL PROTECTED]>, [EMAIL PROTECTED] >(Herman Rubin) wrote: >> >AFAIK there is general agreement that unbiased humans are better at >> >identifying the difference between unpunched holes and imperfectly >> >punched holes than current counting machines -- which after all were >> >only designed to distinguish between unpunched and perfectly punched >> >holes. >> UNBIASED humans, yes. There is considerable evidence of bias. >> Some of this bias is inadvertent, the type of observer bias >> found in many experimental situations in other fields. This >> is especially the case if it is not merely a piece of hanging >> chad, but a dimple. It also occurs if there is a question >> of multiple voting for an office. >Both the United Kingdom and Germany use the old fashioned >piece-of-paper-take-a-pencil-mark-your-candidate method, and the papers >are always handcounted. >I don't think there is ever any question about "voter intention" unless a >voter deliberately chooses to make his ballot paper undecidable. Just >because a system is oldfashioned doesn't mean it can't be better anyway. Having used them, it is not that hard for an ignorant person to do so. It is also not that hard for a corrupt "counter" to invalidate such a ballot. However, I agree that it is better than the current punched card ballots; I seem to be in the far less than one percent of the voters who uses the rather cumbersome and time-consuming method to check my ballot. The United States used hand counted paper ballots universally until late in the 19th century, and there is no place where it is not legal now. This was even the case when there were dozens (literally) of positions voted upon on the same ballot. The major problem with such ballots is the possibility for election fraud. Possibly this is rare in those countries, but I would not be THAT sure. Mainly mechanical voting machines were introduced primarily to reduce fraud. My off-hand guess as to the number of illegally cast ballots in this election is in the neighborhood of 1,000,000. -- This address is for information only. I do not claim that these views are those of the Statistics Department or of Purdue University. Herman Rubin, Dept. of Statistics, Purdue Univ., West Lafayette IN47907-1399 [EMAIL PROTECTED] Phone: (765)494-6054 FAX: (765)494-0558 = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: "reasonable probability"
In article <90k3vl$[EMAIL PROTECTED]>, [EMAIL PROTECTED] (Herman Rubin) wrote: > >AFAIK there is general agreement that unbiased humans are better at > >identifying the difference between unpunched holes and imperfectly > >punched holes than current counting machines -- which after all were > >only designed to distinguish between unpunched and perfectly punched > >holes. > > UNBIASED humans, yes. There is considerable evidence of bias. > > Some of this bias is inadvertent, the type of observer bias > found in many experimental situations in other fields. This > is especially the case if it is not merely a piece of hanging > chad, but a dimple. It also occurs if there is a question > of multiple voting for an office. Both the United Kingdom and Germany use the old fashioned piece-of-paper-take-a-pencil-mark-your-candidate method, and the papers are always handcounted. I don't think there is ever any question about "voter intention" unless a voter deliberately chooses to make his ballot paper undecidable. Just because a system is oldfashioned doesn't mean it can't be better anyway. = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: "reasonable probability"
In article <[EMAIL PROTECTED]>, P.G.Hamer <[EMAIL PROTECTED]> wrote: >Fred Galvin wrote: .. >US voters are indeed totally free not to vote for a presidential candidate. >Although I suspect that the percentage of the US voting population who >actually choose this option is quite low. >However there is a known failure mode of automatic vote counting systems. >If the voting machines sometimes imperfectly punch holes, the still-attached >chad can obscure the hole on a subsequent machine counting. In which case >the machine count may not accurately represent to intent of the voter. >AFAIK there is general agreement that unbiased humans are better at >identifying the difference between unpunched holes and imperfectly >punched holes than current counting machines -- which after all were >only designed to distinguish between unpunched and perfectly punched >holes. UNBIASED humans, yes. There is considerable evidence of bias. Some of this bias is inadvertent, the type of observer bias found in many experimental situations in other fields. This is especially the case if it is not merely a piece of hanging chad, but a dimple. It also occurs if there is a question of multiple voting for an office. -- This address is for information only. I do not claim that these views are those of the Statistics Department or of Purdue University. Herman Rubin, Dept. of Statistics, Purdue Univ., West Lafayette IN47907-1399 [EMAIL PROTECTED] Phone: (765)494-6054 FAX: (765)494-0558 = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: "reasonable probability"
On Tue, 5 Dec 2000 03:02:29 -0600, Fred Galvin <[EMAIL PROTECTED]> wrote: > On Tue, 5 Dec 2000 [EMAIL PROTECTED] wrote: > > > The message below is at: > > > > http://www.deja.com/[ST_rn=ps]/threadmsg_if.xp?AN=701156345&fmt=text > > > > Legal sophistry is truly amazing, especially the use of the phrase > > "reasonable probability." This case has absolutely nothing to do > > with statistics or with probability. It is a simple case of > > arithmetic--look at the undervote ballots. > > There is nothing *wrong* with "undervote" ballots. Voters are not > required to vote on every office and every question on the ballot. > Probably, *most* people who vote don't fill out their ballots > completely. Voters who choose not to vote for any of the candidates > for president have just as much right to have their vote counted as > anybody else, < break > - what Fred posts is logical up to here ... > and it is *wrong* to take their votes and donate them to > one or another of the candidates. - and here, it is a paranoid extension of logic. > No doubt, many voters make mistakes, > but there is no particular reason to think the *undervotes* are > mistakes:< snip, rest > Fred, you are not reading the news, or you are trolling, or you really need to take a deep breath and divorce yourself from the specious appeals of the pure liars fronting for Bush. Undercount happens. Particularly on punch-machines. Nationwide. George W. Bush signed a reasonable law in 1997 for recovering some of it. (But of course, not all of it is recoverable, and nobody pretends that it is.) There is a little more undercount when there is an unusual rule, as in Florida, whereby the ballots are not "groomed" before machine-counting (except in the 6 Republican-voting counties which seemingly were allowed their own interpretation of state law, and who did that in the first count). Then there is the whole dimpled-chad business, where you seem to have missed the testimony that there's often no dispute about the intentions. Now, as statisticians, I suppose we could support that your implicit notion that *every* ballot should be placed into one of two piles; but your post is the first time I have ever seen that suggested. -- Rich Ulrich, [EMAIL PROTECTED] http://www.pitt.edu/~wpilib/index.html = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: "reasonable probability"
Fred Galvin wrote: > There is nothing *wrong* with "undervote" ballots. Voters are not > required to vote on every office and every question on the ballot. > Probably, *most* people who vote don't fill out their ballots > completely. Voters who choose not to vote for any of the candidates > for president have just as much right to have their vote counted as > anybody else, and it is *wrong* to take their votes and donate them to > one or another of the candidates. No doubt, many voters make mistakes, > but there is no particular reason to think the *undervotes* are > mistakes: you could just as well say that lots of people who punched > the Gore hole really meant to vote for one of the other candidates. US voters are indeed totally free not to vote for a presidential candidate. Although I suspect that the percentage of the US voting population who actually choose this option is quite low. However there is a known failure mode of automatic vote counting systems. If the voting machines sometimes imperfectly punch holes, the still-attached chad can obscure the hole on a subsequent machine counting. In which case the machine count may not accurately represent to intent of the voter. AFAIK there is general agreement that unbiased humans are better at identifying the difference between unpunched holes and imperfectly punched holes than current counting machines -- which after all were only designed to distinguish between unpunched and perfectly punched holes. Peter If you are saying that recounts should recount all votes, and be repeated until the result is clear, I'd have to agree. In the UK we manage to do this within hours -- a day at most -- and tend to think that it's self-evidently the appropriate thing to do. (A recount happens if any of the candidates ask for it. I don't know what the rules are to avoid infinite loops. In practice the count seems to converge quite rapidly and to the technical satisfaction of all candidates.) But UK and US elections are so different that this is definitely not intended as a criticism of US practice. In the UK the count is only for a single constituency and there is only one vote recorded on each ballot paper. But again -- if the result is really close --it eventually boils down to humans examining in detail the relatively small number of papers where the voting intention is a matter of judgement. Perhaps the most telling difference is that our manually based system ensures that there are adequate numbers of trusted bodies about to fully identify and examine questionable votes on a short timescale. So it's a fortunate side-effect rather than an intended feature. = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: "reasonable probability"
I specifically did not vote for some offices. I don't need the "psychic friends" counters to guess how I wanted to vote. I refused to cast a vote in these cases. It's not an "under vote" or an incorrect vote. It's blank. = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: "reasonable probability"
If you are interested, I posted some factual information about undervote ballots under the thread "Legal statistical flimflam." DR In article <[EMAIL PROTECTED]>, Fred Galvin <[EMAIL PROTECTED]> wrote: > On Tue, 5 Dec 2000 [EMAIL PROTECTED] wrote: > > > The message below is at: > > > > http://www.deja.com/[ST_rn=ps]/threadmsg_if.xp?AN=701156345&fmt=text > > > > Legal sophistry is truly amazing, especially the use of the phrase > > "reasonable probability." This case has absolutely nothing to do > > with statistics or with probability. It is a simple case of > > arithmetic--look at the undervote ballots. > > There is nothing *wrong* with "undervote" ballots. Voters are not > required to vote on every office and every question on the ballot. > Probably, *most* people who vote don't fill out their ballots > completely. Voters who choose not to vote for any of the candidates > for president have just as much right to have their vote counted as > anybody else, and it is *wrong* to take their votes and donate them to > one or another of the candidates. No doubt, many voters make mistakes, > but there is no particular reason to think the *undervotes* are > mistakes: you could just as well say that lots of people who punched > the Gore hole really meant to vote for one of the other candidates. > > > If this is not done by Florida officials, it will be done by the > > media under the FOI law. Unless these ballots are destroyed, we > > will know the true results. It's only a matter of time. > > How are we going to do that, have psychics handle the ballots? That > might work. The problem is, if they are really psychic, they would be > able to tell who cast the vote, and that would violate the secrecy of > the ballot. Sent via Deja.com http://www.deja.com/ Before you buy. = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: "reasonable probability"
On Tue, 5 Dec 2000 [EMAIL PROTECTED] wrote: > The message below is at: > > http://www.deja.com/[ST_rn=ps]/threadmsg_if.xp?AN=701156345&fmt=text > > Legal sophistry is truly amazing, especially the use of the phrase > "reasonable probability." This case has absolutely nothing to do > with statistics or with probability. It is a simple case of > arithmetic--look at the undervote ballots. There is nothing *wrong* with "undervote" ballots. Voters are not required to vote on every office and every question on the ballot. Probably, *most* people who vote don't fill out their ballots completely. Voters who choose not to vote for any of the candidates for president have just as much right to have their vote counted as anybody else, and it is *wrong* to take their votes and donate them to one or another of the candidates. No doubt, many voters make mistakes, but there is no particular reason to think the *undervotes* are mistakes: you could just as well say that lots of people who punched the Gore hole really meant to vote for one of the other candidates. > If this is not done by Florida officials, it will be done by the > media under the FOI law. Unless these ballots are destroyed, we > will know the true results. It's only a matter of time. How are we going to do that, have psychics handle the ballots? That might work. The problem is, if they are really psychic, they would be able to tell who cast the vote, and that would violate the secrecy of the ballot. -- It takes steel balls to play pinball. = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
