Violence/Genocide: Do not condone violence or genocide on a person or group of people. You are free to attack a person or groups ideas but you are crossing the line when calling for violence. This will be heavily enforced in threads with breaking news involving victims.

Social As NBA ratings mysteriously fall, UFC ratings rise.

Discussion in 'The War Room' started by MigitAs, Aug 26, 2020.

  1. Willy Knuckles Brown Belt

    Joined:
    Nov 21, 2019
    Messages:
    4,071
    Likes Received:
    2,134
    Location:
    Portland, OR
    This whole thread is delusional bro. Dont even bother
     
  2. Zankou Bringing peace and love Staff Member Senior Moderator

    Joined:
    May 24, 2004
    Messages:
    31,126
    Likes Received:
    666
    Location:
    Los Angeles
    I’ll make it simple, since I’ve litigated in this area for many years, both attacking and defending top experts in the field.

    There are many many ways a poll can be deficient or have severe bias, causing it to be non-representative.

    That said, *sample size* is almost never the issue. For the mathematically illiterate, it seems very surprising that one can extrapolate from a couple thousand test samples to a population. But from a statistical perspective, that is almost never a problem because the chance of the sample being *non-representative* very quickly becomes vanishingly small.

    A vastly more serious problem, for example, is the source of the samples. To illustrate, if you use a telephone poll, you are potentially introducing significant distortion. That won’t change if you poll 6000 or sixty million.

    ‘Small sample’ is the #1 critique that people normally make when they don’t understand anything about how polls work from a scientific perspective, and are just reacting to them from a ‘common sense intuition’ perspective.
     
    Pseudo Sane likes this.
  3. Zankou Bringing peace and love Staff Member Senior Moderator

    Joined:
    May 24, 2004
    Messages:
    31,126
    Likes Received:
    666
    Location:
    Los Angeles
    Btw, you can run into problems when the samples are genuinely too small. But this is statistically easy to ascertain. It becomes a major problem in social scientific experiments when you have studies based on 20-50 people. Reputable polling organizations, by contrast, use as standard procedure sample sizes that are way over the statistical baselines, and have a defined scope of error for sample size that is very very small. They know math and they follow basic rules when they conduct these things.
     
  4. Willy Knuckles Brown Belt

    Joined:
    Nov 21, 2019
    Messages:
    4,071
    Likes Received:
    2,134
    Location:
    Portland, OR
    But it was never reveled what demographic that interviewed. And when it comes to something like opinions on why they've stopped watching a show, the answers can vary dramatically, so a larger sample size would for sure help. Again you sound so pretentious and ironically criticize me for my "common sense" approach when you seem to using none. 2000 people in this type of poll is not an accurate measuring tool on the opinions of Millons lol
     
  5. Willy Knuckles Brown Belt

    Joined:
    Nov 21, 2019
    Messages:
    4,071
    Likes Received:
    2,134
    Location:
    Portland, OR
    So you can see why 2,000 people for a poll meant to represent a multi million viewer audience might not be the most accurate? Idek why I was challenged on this. Thats way too small of a sample size, that's obvious
     
  6. Zankou Bringing peace and love Staff Member Senior Moderator

    Joined:
    May 24, 2004
    Messages:
    31,126
    Likes Received:
    666
    Location:
    Los Angeles
    Look, I can’t help you at this point. The Cliffs here are that this poll may have significant problems in how it was structured. We don’t know that, without examining the poll structure and response coding in detail. Which would be boring AF btw, I have dealt with it many times.

    Sample size, however, is neither the problem nor the fix for those potential deficiencies. Put another way, poll 100x as many people, it wouldn’t fix such problems and would do almost nothing from a statistical perspective to make it more accurate. Once you get over a certain threshold sample size is almost never the problem. You don’t have to accept this poll as accurate, since polls can and should be critiqued, but you are barking up the wrong tree as to the potential problems.
     
  7. Willy Knuckles Brown Belt

    Joined:
    Nov 21, 2019
    Messages:
    4,071
    Likes Received:
    2,134
    Location:
    Portland, OR
    I guess I gotta agree to disagree. It seems like you're saying sample size is irrelevant except if the sample size is ridiculously small. I'm asserting that 2000 people in this type of poll is a ridiculously small sample size. Seems like the bigger issue is people posting any kind of poll or "stats" they can find as fact, and it goes un challenged or the goal post gets moved instead of calling out the person for presenting these things as fact. Like I said though, I'll just agree to disagree
     
  8. Kandyland Double Yellow Card Double Yellow Card

    Joined:
    Aug 15, 2019
    Messages:
    2,204
    Likes Received:
    3,608
    Location:
    Huk'd on foniks wurk'd for me
    2,000 people out of 330 million people is literally 0.00061% of the population. That is extremely small and almost irrelevant to the population. Regardless of what you're saying I'm going to have to side with @Willy Knuckles on this.

    But I get what you're saying. There's no real definite way of making a poll accurate. If you poll people in LA, you're going to get a different opinion than people polled in Utah.

    Hence the reason people of all sorts need to take polls with a grain of salt.
     
    Willy Knuckles likes this.
  9. Zankou Bringing peace and love Staff Member Senior Moderator

    Joined:
    May 24, 2004
    Messages:
    31,126
    Likes Received:
    666
    Location:
    Los Angeles
    All I can say is look into statistics. To the layman it seems surprising that a small randomly selected segment of a population would be representative. Mathematically, however it’s just basic math. Almost all experimental science is conducted on minuscule sample sets relative to its extrapolation.

    Otherwise every time you ran a study on lab mice you could object ‘but we only tested 600 mice, and there are 600 billion of them.’ As soon as you pass a very low statistical threshold that becomes meaningless. The actual issue is sample selection, not sample size. The same holds true across all experimental design. You are almost always testing only a minute fraction of the total subject that you are extrapolating to. Generally that is not a problem unless you are running extremely tiny sample sizes, like 20-40.

    And again, if you do have a sample selection problem, then taking 50x more of those samples will do fuck all to address or resolve it. It’s a different issue.

    This is all to say that sample size often jumps out at people as the problematic part, when scientifically it is actually very easy to calculate and control for sample size. In polling, your problem is not that 200,000 people would give markedly different results than 2,000. It’s that you are only asking a distorted subset of people, or that your questions are leading, or that you are asking about subjects that are not easily assessed by the subjects (for example, their future intentions, as with voting).
     
  10. HOLA Silver Belt

    Joined:
    Oct 13, 2012
    Messages:
    10,079
    Likes Received:
    500
    Location:
    Teegeeack
    It does seem like too insignificant of a sample size, but it's not. My statistics class back in college was one of the more interesting classes I took. Lots of smart people really good at math have spent a lot of time perfecting this stuff.
     
  11. Bearknuckle Brown Belt

    Joined:
    Aug 16, 2014
    Messages:
    3,599
    Likes Received:
    358
    Theres a difference between a few fighters going to support the president they love, and the entire organization painting BLM on the canvas, openly speaking about it during the broadcasts, wearing names of people on their helmets, wearing SJW slogans on their jerseys, and making a spectacle out of the national anthem for “protests”. If u don’t see which side is trying to brainwash and guilt trip hardwork americans, then ur blind.
     
  12. Revolver you sound poor

    Joined:
    Feb 12, 2008
    Messages:
    43,886
    Likes Received:
    2,150
    Location:
    Awesome

Share This Page

  1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.