From link@alumni.indiana.edu Fri Nov 17 12:43:29 2000 Date: Thu, 16 Nov 2000 02:18:01 -0500 Subject: Re: Fwd: what do you make of this? (fwd) crt- I've expanded my response to this to include a few of my politically active fiends because I'm curious as to what they think about it. Oh, and incidentally, the correct answer to "What do you make of this?" is always, "Well, I can make a hat, or a brooch, or a pterodactyl," followed by screechy pterodactyl noises. First, the background, for those of you just joining us: On 16 Nov 2000, at 14:33, Charles Twardy wrote: > >Here's an interesting statistical analysis of precinct return data for > >Duval county: > > > > http://www.netrinsics.com/Duval/Duval.html > > > >The bottom line (quoted): > > > > These charts clearly illustrate a strong correlation between > > Democratic precincts and missing votes for a presidential ticket. > > Almost all precincts with more than 50% Al Gore votes have more than > > 15% missing presidential votes. Almost all precincts with more than > > 50% George Bush votes have less than 15% missing presidential votes. > > > >Interestingly, there were more valid votes for Senator than for > >President in all precincts, implying voter confusion across the board, > >but again, the discrepancy was far higher in predominantly Democratic > >precincts. > > "The most dangerous of all falsehoods is a slightly distorted truth" --G. C. Lichtenberg "The plural of `anecdote' is not `data'." --Unknown I'm not sure you can draw any conclusion from the statistics given because there are so many scenarios you could construct to fit the facts. It could be that democrats are having trouble marking their ballots. It could also be that some of the precincts in Jacksonville that went democrat did so because a large number of republicans botched their ballots. It could be that a large number of voters in Jacksonville had no strong preference between Bush and Gore, and chose to abstain by invalidating that part of their ballots. And, finally, it could be a statistical fluke. All of the interpretations that favor one or another of the candidates no doubt have their advocates, so I'll take it upon myself to defend the last option, since it seems to be otherwise left out in the cold. We all have heard the expression "Correlation does not imply causation," but that expression has an ugly step-sibling that does not often get talked about, "Correlation does not imply significance." There are 67 counties in Florida alone, and probably a few thousand in the nation as a whole. The results in any one county make sense only in the context of similar results from the rest of the counties. For instance, if we were to find that nationwide one county in a hundred showed a correlation like the one in Jacksonville, we should conclude that there is nothing particularly unusual in the Jacksonville results; Duval county would be just an outlier in the distribution of voting behavior in US counties. The important thing to remember is that correlations, even strong correlations can arise by chance. Or, to put it another way, even random noise has patterns in it; in fact, if you see a data set that is devoid of patterns, concentrations, and voids, then it's a pretty safe bet that the data is anything *but* random. Statistics types like to call this the "Texas Sharpshooter Effect". However, even if a comparison to counties in the rest of the US shows that the Jacksonville results are somehow pathological, we still have to consider the possibility that the reason is that Jacksonville itself is somehow pathological. When you get right down to it, no place else is quite like Jacksonville (this probably comes as a relief to the those readers who have been there). It may just be that Jacksonvillains, for reasons known only to themselves, simply vote differently than normal human beings, in which case the only valid comparison population is Jacksonvillains themselves. So, at this point I have made it sound like it's impossible to draw any conclusions whatsoever from any election results from any county, but that isn't my intent. What I am trying to impress on anybody who hasn't already deleted this message by now is that a careful and thoughtful analysis of these results is very difficult; it's not something you just knock off in your spare time during the week following the election, and it's certainly not something you do in a vacuum, without reference to results from other places and in other elections. Make no mistake, most of the people pushing these analysis have a favorite side in this contest, and that subtly colors their judgment, particularly now, while the results of the contest are still in doubt. It seems to me that another thing we need to remember is that an election, like any other measurement, is noisy. It's easy to forget that because we usually think of noise as a property of continuum systems and as something that discrete systems are immune to. Nevertheless there are random factors that affect vote totals. Parties' attempts to get out their vote meet with varying levels of success. Good or bad weather enhances or suppresses turnout. Economic and geopolitical factors sway public sentiment, and, yes, it seems that voters even make errors in the voting booth. For the most part we haven't noticed this noise because in recent history the signal-to-noise ratio of our elections has been remarkably high. I think that is to our credit; we do a good job of keeping the noise in our elections low. I suspect you would be hard pressed to find a better example. I believe that the SNR of this election is no lower than it has been in recent history. Instead, what the results are telling us is that, to within the noise, the election is a tie. When that happens there is no way around it; the noise will determine the winner. Whichever of the candidates wins this election will have to reconcile himself to the fact that he was chosen pretty much at random. More importantly, we will have to reconcile ourselves to that fact. It sucks, perhaps, but it is a fact of life, like death and taxes, and garbage trucks that make that beep, beep sound right outside your open window at 5 in the morning. We can but hope that somewhere out there is a real statesman or stateswoman that is drawing inspiration from the nation's polarization to provide us with the leadership that is so egregiously lacking in contemporary politics. In the meantime, I don't worry too much about who actually wins this election. I worry instead, about the long-term effects that will color future elections. There has been much talk of making the vote counting process in this election "fair," but I see no way that ad-hoc counting rules determined after the margin of victory is known can be "fair" in any meaningful sense of the word. Once you know how many votes you have to make up, your target number, as it were, your sense of fairness is unavoidably tainted by that knowledge. The only way to have a fair process is to spell out the rules before any votes are cast, and limit the decision making after the vote to strictly objective criteria, such as "Is the margin of victory below some threshold," and the like. The precedents we set here will guide the resolution of future disputes, and I fear we are making a muddle of it. I also fear for the Electoral College. Already the demagogues are beating the drums about the possibility of the electoral vote going opposite the popular vote. Worse, there are whispers of pressuring the electors to defect on their votes if the electoral vote is close. I think the system can survive the demagogues, but a defection that changes the result of the election will probably spell the death knell for the Electoral College. At times like this we would do well to bear in mind that government of the people, by the people, for the people does not necessarily mean government of 51% of the people, by 51% of the people, for 51% of the people. The Great Compromise, the separation of powers, and even the Electoral College are all mechanisms put in place to make sure that our Republic was governed by consent of all the governed, not just by 51% of the governed. But maybe that's a rant for another day. Sorry for going on so long; it's been that kind of day. Regards, -rpl