mynameisalex
Rookie
- Nov 5, 2013
- 3
- 0
- 1
Today I was pondering the upcoming elections and my past experiences with them, and I stumbled on an idea. Now I figured I'd post it here just to see the reaction...
This stems from me being skeptical about how informed we all are about the complex issues of governing and which candidates may do the best job.
Before I explain the idea, you might want to know of some basic statistics. Particularly, the idea of random sampling, which, when used correctly, is capable of creating a small population that is surprisingly well representative of a much larger one. These techniques are of course used in sampling polls, which historically have sometimes been far off (generally due to sampling biases), but have improved fairly greatly over time. Random sampling is also used in jury duty selection, and the statistical principles that allow it to be accurate are the same ones used in essentially all of the science and medicine experiments done in these days, to prove whether some technique truly did have some effect.
Now, say I take a state with about 5 million people, and choose a random sample of 5000. It is very likely that that sample will be a good representation of the views of the entire population. Additionally, the statistics also show that increasing the sample size cannot actually improve it much. So a random sample of 100,000 or even 1 million of the people would be only very slightly better (and still extremely close to a complete representation)
Perhaps you see where I am going with this...
Basically, as a thought experiment, rather than having everyone vote, what if we just take a good random sample and just allow those people to vote. Now, before anyone goes berserk on me, let's look at some arguments, and I'll explain a bit more.
First, there is the expected critique: can this really be a good representation?
Well first of all, we would all like to think that our current system gives a great representation, but not necessarily. Voter turnout can be low, and I don't think it is only the people who don't care about the vote who are not voting. To truly vote well, people need to do their research on the candidates and to, of course, make it to the voting places. But many people, including those who are less well off and may not have the time to properly vote, even if it could be important for them to do so, won't get a chance. However, with the random sampling idea there's a solution to this. Similar to jury duty, and since the number of people in the whole sample is small compared to the population, special attention can be taken so that they can vote properly. They can be paid a reasonable amount for their lost time, ensured that they can get to the voting places, etc.
Perhaps even better, this sample population could be given all the resources and time they need to do in-depth analyses of all the candidates (not the superficial stuff many of us are getting). And this could become something like jury duty, for which employers and others understand that the selected person (if they accept) has a responsibility to be thorough and make a truly good selection, for the better of the population. Note though, that unlike jury duty, each individual would not meet with the others or be in any way influenced by their choices.
Plus, there are a few other pros:
-The rest of us wouldn't have to spend time on elections.
-All the election systems (organization, counting votes, etc.) would become much simpler.
-We wouldn't have to see all the dumb political commercials during big elections, or have so much money wasted on those political campaigns (there might still be a little campaigning in the form of presenting information to sample populations but it would be more in depth and on a much smaller scale)
-We might actually get intelligent politicians instead of random people who have good-sounding names, and good looks, and lots of money to waste.
Something else I might add, is that the ballot would be better if it was different than just selecting candidates. For instance, allow the voters to rate every one of the candidates on a scale, and then all the points are added for each candidate. If someone has a very strong preference then they can give one candidate many more points than another, while someone with a weak preference can give them a slightly different number of points; or something similar, I haven't put a great deal of thought into it.
Additionally, I want to add something regarding the accuracy of sampling polls (and thus random sampling). There is a complete distinction between representing a population accurately and predicting the future. A sampling poll might give a great representation of the population (and with correct random sampling the statistics can almost guarantee that it will), but that doesn't mean this can predict what people will do on election day. For one, people might not take the poll as seriously as the actual election, and give a different answer; and also, people change their minds quite a bit. Approval ratings can vary greatly simply due to the approval of the population actually varying over time, and in the end the person who gets elected is whoever has the approval in their favor when election day comes. Something one might take from this is that voting is not always a precise science. Personally, I wouldn't be too surprised if, having given the same exact election twice (repeated all the voting), the results could come out significantly different (due to people changing their votes).
Anyway... what do you think?
This stems from me being skeptical about how informed we all are about the complex issues of governing and which candidates may do the best job.
Before I explain the idea, you might want to know of some basic statistics. Particularly, the idea of random sampling, which, when used correctly, is capable of creating a small population that is surprisingly well representative of a much larger one. These techniques are of course used in sampling polls, which historically have sometimes been far off (generally due to sampling biases), but have improved fairly greatly over time. Random sampling is also used in jury duty selection, and the statistical principles that allow it to be accurate are the same ones used in essentially all of the science and medicine experiments done in these days, to prove whether some technique truly did have some effect.
Now, say I take a state with about 5 million people, and choose a random sample of 5000. It is very likely that that sample will be a good representation of the views of the entire population. Additionally, the statistics also show that increasing the sample size cannot actually improve it much. So a random sample of 100,000 or even 1 million of the people would be only very slightly better (and still extremely close to a complete representation)
Perhaps you see where I am going with this...
Basically, as a thought experiment, rather than having everyone vote, what if we just take a good random sample and just allow those people to vote. Now, before anyone goes berserk on me, let's look at some arguments, and I'll explain a bit more.
First, there is the expected critique: can this really be a good representation?
Well first of all, we would all like to think that our current system gives a great representation, but not necessarily. Voter turnout can be low, and I don't think it is only the people who don't care about the vote who are not voting. To truly vote well, people need to do their research on the candidates and to, of course, make it to the voting places. But many people, including those who are less well off and may not have the time to properly vote, even if it could be important for them to do so, won't get a chance. However, with the random sampling idea there's a solution to this. Similar to jury duty, and since the number of people in the whole sample is small compared to the population, special attention can be taken so that they can vote properly. They can be paid a reasonable amount for their lost time, ensured that they can get to the voting places, etc.
Perhaps even better, this sample population could be given all the resources and time they need to do in-depth analyses of all the candidates (not the superficial stuff many of us are getting). And this could become something like jury duty, for which employers and others understand that the selected person (if they accept) has a responsibility to be thorough and make a truly good selection, for the better of the population. Note though, that unlike jury duty, each individual would not meet with the others or be in any way influenced by their choices.
Plus, there are a few other pros:
-The rest of us wouldn't have to spend time on elections.
-All the election systems (organization, counting votes, etc.) would become much simpler.
-We wouldn't have to see all the dumb political commercials during big elections, or have so much money wasted on those political campaigns (there might still be a little campaigning in the form of presenting information to sample populations but it would be more in depth and on a much smaller scale)
-We might actually get intelligent politicians instead of random people who have good-sounding names, and good looks, and lots of money to waste.
Something else I might add, is that the ballot would be better if it was different than just selecting candidates. For instance, allow the voters to rate every one of the candidates on a scale, and then all the points are added for each candidate. If someone has a very strong preference then they can give one candidate many more points than another, while someone with a weak preference can give them a slightly different number of points; or something similar, I haven't put a great deal of thought into it.
Additionally, I want to add something regarding the accuracy of sampling polls (and thus random sampling). There is a complete distinction between representing a population accurately and predicting the future. A sampling poll might give a great representation of the population (and with correct random sampling the statistics can almost guarantee that it will), but that doesn't mean this can predict what people will do on election day. For one, people might not take the poll as seriously as the actual election, and give a different answer; and also, people change their minds quite a bit. Approval ratings can vary greatly simply due to the approval of the population actually varying over time, and in the end the person who gets elected is whoever has the approval in their favor when election day comes. Something one might take from this is that voting is not always a precise science. Personally, I wouldn't be too surprised if, having given the same exact election twice (repeated all the voting), the results could come out significantly different (due to people changing their votes).
Anyway... what do you think?
Last edited: