Opinion My opinion on H.R. 1186 "Keep Americans Safe Act" (high cap. mag. ban) LONG READ

[rambosabeast]

That's actually not a bad sample size. The point of a sample is that its, well, just a sample from a population so it doesn't need to come anywhere near the size of the population its drawn from. As long as its representative and over a certain size its a fine sample. In this case we're talking about a population of ~150,000,000 so for a 95% confidence level with a confidence interval of 2(what the source claims it has) the ideal sample size would be ~2400. So its not perfect since its a little short but its not like its BS either.

I'm not really familiar with confidence intervals, so I'll take your word for that. But I disagree that a voluntary online survey (convenience sampling) which samples .0013% of a population (Republican voters) can be used to draw any meaningful conclusions. There is also another potential flaw in the methodology which could result from respondents claiming to be Republican voters in order to influence a narrative. Morning Consult can't verify the respondents honesty. That's what people are responding to anyway. If CNN does a poll, the respondents aren't responding to CNN, they're responding to their opinion of CNN and gauging how their answers affect CNN's position or biases. Same goes for the gun control issue. People will respond in a way that they believe will give credence to their "team". So not to discount all political polling, I would argue that any claims made about that data shouldn't be made in the form of "X% of voters say this" when there is no statistical or practical significance to the results in comparison to the population.

 

[Islam Imamate]

I'm not really familiar with confidence intervals, so I'll take your word for that. But I disagree that a voluntary online survey (convenience sampling) which samples .0013% of a population (Republican voters) can be used to draw any meaningful conclusions.

The fact that the sample is only just that, a sample, of the target population isn't a problem in the slightest. With this logic we can't conduct any surveys since its virtually impossible for most organizations to survey a chunk of the entire population they're looking at. And you don't even need that, like I said for a population as large as that of the registered voting populace(~150,000,000) you only need a sample size of ~2400 as long as the sample is representative.

There is also another potential flaw in the methodology which could result from respondents claiming to be Republican voters in order to influence a narrative. Morning Consult can't verify the respondents honesty. That's what people are responding to anyway. If CNN does a poll, the respondents aren't responding to CNN, they're responding to their opinion of CNN and gauging how their answers affect CNN's position or biases. Same goes for the gun control issue. People will respond in a way that they believe will give credence to their "team". So not to discount all political polling, I would argue that any claims made about that data shouldn't be made in the form of "X% of voters say this" when there is no statistical or practical significance to the results in comparison to the population.

That's why you take large samples, to diminish the effect of those kinds of respondents. Its also why you include the margin of error(in that survey its 2%). No offense but this is the kind of argument people make when they don't like the results of a survey. Not saying Morning Consult has perfect methodology but these specific criticisms don't hold much water IMO.

And I say this as someone who completely disagrees with an AWB. So personally I don't like the results since I think those voters are wrong and an AWB is not good policy.

 

[rambosabeast]

The fact that the sample is only just that, a sample, of the target population isn't a problem in the slightest. With this logic we can't conduct any surveys since its virtually impossible for most organizations to survey a chunk of the entire population they're looking at. And you don't even need that, like I said for a population as large as that of the registered voting populace(~150,000,000) you only need a sample size of ~2400 as long as the sample is representative.

That's why you take large samples, to diminish the effect of those kinds of respondents. Its also why you include the margin of error(in that survey its 2%). No offense but this is the kind of argument people make when they don't like the results of a survey. Not saying Morning Consult has perfect methodology but these specific criticisms don't hold much water IMO.

And I say this as someone who completely disagrees with an AWB. So personally I don't like the results since I think those voters are wrong and an AWB is not good policy.

Well to be fair to me, I'm not dissatisfied with the polling responses per se, I'm only arguing that based on my understanding of statistics and polling, there are flaws which detract from the poll. The sample is not representative of the population due to its size, and the methodology/design make the results less meaningful. I'd argue the same points if the polling question was anything else.

Why does a sample size of 2,400 allow us to draw conclusions about a population of ~150,000,000?

Could I safely assert that the majority of Democratic voters prefer liqour over beer, when I only sampled .0013% of them? It seems irresponsible to make such a claim with such small data.

 

[rambosabeast]

@Kafir-kun I think of it this way: Would a politician look at this data and truly believe that there's a paradigm shift taking place regarding AWB within the party's base? With good faith and an unbiased interpretation, I would say the answer is no. Perhaps I am misunderstanding statistical data but that's basically my position.

 

[Islam Imamate]

Well to be fair to me, I'm not dissatisfied with the polling responses per se, I'm only arguing that based on my understanding of statistics and polling, there are flaws which detract from the poll. The sample is not representative of the population due to its size, and the methodology/design make the results less meaningful. I'd argue the same points if the polling question was anything else.

Why does a sample size of 2,400 allow us to draw conclusions about a population of ~150,000,000?

Could I safely assert that the majority of Democratic voters prefer liqour over beer, when I only sampled .0013% of them? It seems irresponsible to make such a claim with such small data.

This is a really common misconception. There's nothing wrong with drawing data from samples like that, its pretty much the only way to conduct surveys since surveying the entire population isn't really feasible. As long as the sample is representative its fine. When I say that I mean that if ~20% of the voting population is over 55 then ~20% of your sample should be over 55 and if ~10% of the population is black then ~10% of your sample should be black. There's no need to survey the entire population to conduct surveys like this, that's a pretty unreasonable bar to set.

@Kafir-kun I think of it this way: Would a politician look at this data and truly believe that there's a paradigm shift taking place regarding AWB within the party's base? With good faith and an unbiased interpretation, I would say the answer is no. Perhaps I am misunderstanding statistical data but that's basically my position.

I think you are misunderstanding statistics.