I do believe that a sample size of 1,000 is accurate.
/thread?
/yourself?
:ugh:
Why can we not discuss these matters to further length? I would love to go into deeper detail.
I found a great article for you. Please read it. I’m reading it right now and I find it pretty interesting.
How Are Polls Conducted?
jake, i see your point… but just admit to yourself that polls are just ratios or percentages of the total public opinion… a 100% ‘accurate’ poll is when the sample size = population… so any deviation of that, results in a ratio with a margin of error associated with it. so there’s your inaccuracy in polling.
/thread
That is not the point being argued. I have always talked in these threads about a margin of error and confidence levels. The point that is currently up for debate is that 1,000 randomly selected samples can represent the population as a whole. Due to the exponential form of confidence levels and interval selection, as you increase your sample size your jump in confidence decreases exponentially. An example used in the article that I just posted mentions that if they poll 1,000 people, there is a 95% confidence level. If they double the size of the poll to 2,000 people, there is only a 2% increase in the confidence level. For statistics to be realiable confidence levels need to meet or exceed 95%.
And I do completly agree that statistics are a % of the national population. That is what statistics is all about.
x2
…and I read a similar article already, and I don’t find it interesting. In the end its still a math equation with them using a severly small fraction of people to “guess” what the rest of the country is thinking.
and by guess you mean selecting an accurate representation of the population?
Just because their sample consists of a identical percent of racial, gender, and/or age coinsistant of the country doesn’t mean that everyone thinks alike. Christ thats as bad as saying all blacks are degenerates because their black so they act the same way…
Lets take it to another topic. You ask 1000 women (age/race) if they believe in abortion. You use a computerized phone list and poll away. Who is not to say that you happen to get a over wheliming responce swayed one way or the other just by who is on the other end of the phone. Then you take your findings and are going to try and say “well we asked 1000 women with a % make up of that equal to the country and they say they are XXX on abortion” then it cracks me up because the way its always presentd is “from a recent poll conducted, the country feels”…Um no those 1000 people (or however many in the poll) feel not the country.
That prove statistics right there. If polling is random (and in no way based on racial, gender, and age) and the sample has identical percentages of race, gender, and age that are consistant to the country, then that means that it must be an accurate representation of the population.
Lets take it to another topic. You ask 1000 women (age/race) if they believe in abortion. You use a computerized phone list and poll away. Who is not to say that you happen to get a over wheliming responce swayed one way or the other just by who is on the other end of the phone.
the statisticians who crunch the numbers and determine an appropriate sample size are to say that you will not get an overwhelming response swayed one way or the other. You can calculate the numbers if you want and determine the odds of this happening. If we had a small sample size, such as 10, a bias in the sample likely to occur. This is why you have a large sample size to remove this bias.
Then you take your findings and are going to try and say “well we asked 1000 women with a % make up of that equal to the country and they say they are XXX on abortion” then it cracks me up because the way its always presentd is “from a recent poll conducted, the country feels”…Um no those 1000 people (or however many in the poll) feel not the country.
Let me use an example for you. On September 21, 2001, Bush’s presidential approval rating was at 90%. Do you believe that this was ‘made up’ and did not represent the nation as a whole?
I don’t undersatand what you are trying to say (maybe its a typo or wording). From what I read with Polling inorder for the highest accuracy it can;t just be random. It has to have a set number of and it has to have a race/gender/age % which is equal to the country.
the statisticians who crunch the numbers and determine an appropriate sample size are to say that you will not get an overwhelming response swayed one way or the other. You can calculate the numbers if you want and determine the odds of this happening. If we had a small sample size, such as 10, a bias in the sample likely to occur. This is why you have a large sample size to remove this bias.
if its not 100% there is no way of knowing what the true opinon is. I don’t care if its 10 people from 300 million or 100,000 out of 300 million.
Let me use an example for you. On September 21, 2001, Bush’s presidential approval rating was at 90%. Do you believe that this was ‘made up’ and did not represent the nation as a whole?
No I don’t believe it represents the nation as a whole. Out of 300 million, only 122,229,000 people that were elligible or cared enough to vote, did and thats not even half of the country.
I believe that polls are unprecise no matter how well executed or well manored they are ment to be.
For polling to be unbiased, it must be random. That random sample, if done correctly should be equal to race/gender/age % of the population as a whole. Let’s take race as an example. Say the national African-American population is 15%. Picking a random phone number to poll someone, I have a 15% chance of calling an African-American. If Caucasians make up 70% of the national population, I have a 70% chance of calling a Caucasian. So once my total sample has been taken, I should have basically the same percentages. If I call 1,000 people, 700 should be Caucasian and 150 should be African American. The equal representation was based due to the sheer randomness of it.
I’ll give you an example of a biased poll. In 1936, Literary Digest used a mail in sheet to try and determine who would win the election. Over 2 million (!) people responded. This was the first year that Gallup started doing presidential elections. He took a sample of 1,500 people. Who ended up being correct? Gallup, with a sample size of 1,500. The reason is because Literary Digest had a biased sample, less lower class workers subscribed to it, thus neglecting part of the population. Gallups poll was random, ensuring that it was an accurate representation of the population due to its randomness.
if its not 100% there is no way of knowing what the true opinon is. I don’t care if its 10 people from 300 million or 100,000 out of 300 million.
That is obviously true, but the question is what is an acceptable level? Surely polling 300 million people is not feasible. But due to statistical measuring, we can have an answer that is has a 5% chance of error with a much smaller sample size.
No I don’t believe it represents the nation as a whole. Out of 300 million, only 122,229,000 people that were elligible or cared enough to vote, did and thats not even half of the country.
The data could easly be represented to narrow it to only those who voted. I am sure that in these polls the question is also raised if they are an active voter. You can sign up at gallup.com for $95 a year if you are really interested.
I believe that polls are unprecise no matter how well executed or well manored they are ment to be.
What is your definition of precise? What do you feel is an acceptable confidence level?
Look in all seriousness, I’m not playing into this “keep the thread going game” because I have lost interest as stated in the past couple of my posts. The key word in all of this is “acceptable”. I do not accept the polls because they can not be 100% correct, I’m sorry if this pisses you off. I’m happy you are in college and learning about this while finding it all facinating, I don’t. I stated my argument and explained myself to you for whatever reason and I’m done. You can take your “i’m in college and your not innuendos” and shove them.
Hope thats not too point blank for you to get offended.
What the fuck are you arguing about? It is flawed due to statistical errors… you just said so yourelf.
May not be alot (like you said) but there is an understanding that there may or may not be a 6 percent error. (+ and - 3%)
I’m not offended at all. I just find it amazing for you to be able to make such a broad generalization saying that, “All polls are flawed.”
I have no problem with you not being in college. Neither of my parents went to college and they’re some of the people I respect the most. My problem is with you making a statement such as “All polls are flawed!” and then not willing to delve deeper and say why you believe this.
One question that i want to ask and then i’ll leave you alone is this:
What do you think the presidential approval rating is? If this poll is flawed, what do you believe are the true numbers?
I think these polls can tell you a lot about the state of how people feel. To say “All polls are flawed” (to the point where they’re not even worth looking at) is absurd. These polls help to show how the presidents various policies affect the general american views. As I said before, Bush’s approval rating was at 90% the week following 9/11. If these polls were ‘flawed’ then they should not have jumped at all and remained at pre-9/11 levels (even though i think we can safely assume that the vast majority of americans felt Bush’s standpoint the week after 9/11 was approriate).
I never said I didnt look at them. I just think its stupid to base everything off them… while viewing poll results one should keep in mind it could be stacked, or flawed in some ways.
Now can we agree?
agreed… sometimes i get in arguments just for the sake of arguing :embarassd
I got no work done this morning at work just because of this thread :embarassd
I agrue lots too… haha… pittspeed is the best place to argue… keeps me from doing it in real life. haha