Do you dispute the claims in this Gelman paper about the probability of votes in various states being decisive in Presidential elections? Or the much higher probabilities of decisive individual impact in state and local races, and popular referenda/initiatives?
Lotteries are for personal consumption, and have negative expected value. Voting can be done as an act of altruism (in addition to other reasons), buying a small chance of very large impact, for which it can easily have a positive expected value. It would cost hundreds of dollars in political contributions at least to pay for the delivery of another vote to replace yours, so there is a large wedge between your opportunity cost of time and your productivity voting.
Voting can be done as an act of altruism (in addition to other reasons), buying a small chance of very large impact, for which it can easily have a positive expected value.
I agree with this argument, but note that it only applies if you actually believe that the differences between the policies that Obama and Romney are likely to implement do amount to a very large overall utility differential (and that you can know beforehand in which direction it goes). I suspect that Konkvistador does not share this premise.
Large enough to be millions of times what you would buy with a $50 charitable donation. That’s not a terribly high bar for differences between candidates. And certainly one’s vote is more influential in primary elections than general elections, and in swing states, and in lower turnout regions, etc. Policy differences can also be clearer in other races and cases, e.g. voting on single initiatives in California.
The paper assumes votes are accurately recorded, counted, and reported. Which is known to be false; error rates in vote counts are at least 0.1%, and likely closer to 1%. A perfectly honest close election is an election decided not by actual votes cast, but the random distribution of counting errors. And any election so close is going to be subjected to recounts that simply redistribute the counting errors.
Now, it is theoretically possible your vote might actually tip things in the final recount, right? Despite the fact that who actually won in a close election is unknown and unknowable, your vote is more likely to be accurately counted than not, so it might tip over the decision, right?
Except that’s assuming perfect honesty in recording, counting, and reporting, which is ridiculous. What will determine who wins in a close election is whether the margin created by random counting errors is small enough that the people in the best position to commit fraud can tip it the way they prefer.
And, of course, we then ask—did you actually have a good, reliable of idea how your candidate was going to do in office, and then on top of that how his choices were actually going to translate into effects? Really? So, back in November 2008, what did you predict the September 2012 unemployment rate would be, if Obama won? What did you predict the US budget deficit would be? Did you predict that the average number of deaths of US personnel in Afghanistan per month under Obama would be five times higher than it was under Bush? Did you predict the overthrow of the Libyan government by US air power? Let’s be serious; Obama didn’t have a very good idea of how his policies would translate into actual effects back on Election Day 2008.
Your vote for a position less powerful than President is more influential, sure, but its actual effect is reduced because the position is less powerful. There might be some point in voting on propositions and initiatives if your state has them, and maybe on very local elections if you’ve bothered to become informed on them and live in a small enough community.
Assuming honesty in recording is actually not problematic. As Eugine_Nier says, there will still be a set of voting outcomes that lead to Candidate A being elected, and a set of voting outcomes that lead to Candidate B being elected, and fraud only slightly changes the shape of the boundary between these sets.
It gets better. Turns out that the “area” of that boundary is minimized in a fair majority election. The probability of a vote being pivotal is only increased when the boundary is distorted by fraud (although, obviously, your vote will no longer be pivotal in exactly the same situations).
If the error rate in vote counts is 1%, that means you’re 99% as likely to make the vote you intend to make. So if you had a 1 in 10 million chance to make a pivotal vote, that chance now becomes… roughly 1 in 10.1 million. This part doesn’t really make a lot of difference, although you’re right that it should be taken into account.
What does that have to do with anything? Okay, fine, make the error rate 10%. Then your chance of making a pivotal vote just became 1 in 11 million instead of 1 in 10 million. That’s a gross overestimate and it still hasn’t made a huge difference.
Edit: My point is that although dishonesty changes when exactly your vote is pivotal, it increases the probability that it will be.
I have no idea why this post is down voted, since it points out something very important, voting results are an imperfect measurement of who the electorate actually tried to vote for.
Except that’s assuming perfect honesty in recording, counting, and reporting, which is ridiculous. What will determine who wins in a close election is whether the margin created by random counting errors is small enough that the people in the best position to commit fraud can tip it the way they prefer.
Your vote might still be the vote that tips the total past the threshold where the opposing counters can commit fraud.
Do you dispute the claims in this Gelman paper about the probability of votes in various states being decisive in Presidential elections? Or the much higher probabilities of decisive individual impact in state and local races, and popular referenda/initiatives?
Lotteries are for personal consumption, and have negative expected value. Voting can be done as an act of altruism (in addition to other reasons), buying a small chance of very large impact, for which it can easily have a positive expected value. It would cost hundreds of dollars in political contributions at least to pay for the delivery of another vote to replace yours, so there is a large wedge between your opportunity cost of time and your productivity voting.
I agree with this argument, but note that it only applies if you actually believe that the differences between the policies that Obama and Romney are likely to implement do amount to a very large overall utility differential (and that you can know beforehand in which direction it goes). I suspect that Konkvistador does not share this premise.
Large enough to be millions of times what you would buy with a $50 charitable donation. That’s not a terribly high bar for differences between candidates. And certainly one’s vote is more influential in primary elections than general elections, and in swing states, and in lower turnout regions, etc. Policy differences can also be clearer in other races and cases, e.g. voting on single initiatives in California.
The paper assumes votes are accurately recorded, counted, and reported. Which is known to be false; error rates in vote counts are at least 0.1%, and likely closer to 1%. A perfectly honest close election is an election decided not by actual votes cast, but the random distribution of counting errors. And any election so close is going to be subjected to recounts that simply redistribute the counting errors.
Now, it is theoretically possible your vote might actually tip things in the final recount, right? Despite the fact that who actually won in a close election is unknown and unknowable, your vote is more likely to be accurately counted than not, so it might tip over the decision, right?
Except that’s assuming perfect honesty in recording, counting, and reporting, which is ridiculous. What will determine who wins in a close election is whether the margin created by random counting errors is small enough that the people in the best position to commit fraud can tip it the way they prefer.
And, of course, we then ask—did you actually have a good, reliable of idea how your candidate was going to do in office, and then on top of that how his choices were actually going to translate into effects? Really? So, back in November 2008, what did you predict the September 2012 unemployment rate would be, if Obama won? What did you predict the US budget deficit would be? Did you predict that the average number of deaths of US personnel in Afghanistan per month under Obama would be five times higher than it was under Bush? Did you predict the overthrow of the Libyan government by US air power? Let’s be serious; Obama didn’t have a very good idea of how his policies would translate into actual effects back on Election Day 2008.
Your vote for a position less powerful than President is more influential, sure, but its actual effect is reduced because the position is less powerful. There might be some point in voting on propositions and initiatives if your state has them, and maybe on very local elections if you’ve bothered to become informed on them and live in a small enough community.
Assuming honesty in recording is actually not problematic. As Eugine_Nier says, there will still be a set of voting outcomes that lead to Candidate A being elected, and a set of voting outcomes that lead to Candidate B being elected, and fraud only slightly changes the shape of the boundary between these sets.
It gets better. Turns out that the “area” of that boundary is minimized in a fair majority election. The probability of a vote being pivotal is only increased when the boundary is distorted by fraud (although, obviously, your vote will no longer be pivotal in exactly the same situations).
If the error rate in vote counts is 1%, that means you’re 99% as likely to make the vote you intend to make. So if you had a 1 in 10 million chance to make a pivotal vote, that chance now becomes… roughly 1 in 10.1 million. This part doesn’t really make a lot of difference, although you’re right that it should be taken into account.
I don’t think you appreciate just how hard counting votes is.
What does that have to do with anything? Okay, fine, make the error rate 10%. Then your chance of making a pivotal vote just became 1 in 11 million instead of 1 in 10 million. That’s a gross overestimate and it still hasn’t made a huge difference.
Edit: My point is that although dishonesty changes when exactly your vote is pivotal, it increases the probability that it will be.
I have no idea why this post is down voted, since it points out something very important, voting results are an imperfect measurement of who the electorate actually tried to vote for.
Your vote might still be the vote that tips the total past the threshold where the opposing counters can commit fraud.