Liberal media bias? The hell you say.Follow

Well, I took a statistics course in high school where we analyzed the probability of getting a red M&M, so I should probably weigh in, too.





Nah.

Really? Because you missed a simple point and I laughed at you and you've been throwing hissy fits for the last ten posts or so?

Well, if you say so. I certainly feel chastised, let me tell ya.

I'll tell you what: You give me three good guesses as to what you think you missed and I'll tell you. :)

Edited, May 25th 2011 6:04pm by Jophiel

You're being childish gbaji. Can't we at least agree that any poll in which 38% of the respondents self-identify as conservatives and only 23% self-identify as liberals is going to be skewed in favor of the leader of the conservative movement? Yes or no? I didn't think this would be such a hard thing for you to admit, but it's like you have a mental block or something.

So how does that compare to:

38% conservative
36% moderate
23% liberal

In the exact same poll, mind you.

Lol! Um... a 2% difference between the largest two groups, compared to a 17% difference. So I'd say that it doesn't even come close to being as skewed. Let's make a simpler comparison. Let's pretend that magically exactly half of the middle group in each set split for each end point. Just for kicks, where does that put us:

58% Democrat vs 41% Republican

51% Conservative vs 46% Liberal.


One of those things is much more skewed than the other. Can you see which one?

Moderates are the swing voters that go between Republican and Democrat, no?

Why do you keep saying 3 group split as if the correct answer must be that all answers are supposed to have equal weight? I think that's where you're tripping yourself up.

So you're assuming that some of the moderates haven't already done that and, again, they MUST SPLIT EQUALLY? That's a pretty stupid assumption.

Oh wait. Good, looks like that wraps everything up.






Edited, May 25th 2011 7:11pm by bsphil

Not a thing. Which is why I'm confused at how poorly you understand it. I know what you saying, it's just that what you're saying is mired in mistaken impressions and misunderstandings.

Er? Reading phail? Moderates are the group between Liberal and Conservative. You're mixing up the sets.



That's just it though. We don't know what the actual distribution of those groups are in the population. Hence the point of showing the Rasmussen poll in which only 33% of those polled self-identified as Democrats compared to the AP poll which had 46%. Joph's counter to that argument was that it could be that the Rasmussen poll is skewed and not the AP poll. He then followed that up with some argument about how people are more likely to shed their party ID instead of their ideology. When I pointed out that this supported the lower and more balanced numbers in the Rasmussen poll instead of the more skewed numbers in the AP poll, he quickly backpedaled and insisted that he wasn't saying that as any sort of explanation as to why the AP poll numbers must be correct and the Rasmussen numbers must be wrong.


And he then proceeded to refuse to provide any alternative explanation either. So wishful thinking I guess?



It's an assumption which doesn't favor either side. That's the point. It's just an arbitrary means to see whether a value seems over represented. How else would you assess those numbers? Just look at them and go "OMG! 38% is just as much of an over-representation of a group as 46%!!!".

Really? You honestly can't see that 46 is higher than 38? I've seen a lot of dumb on this board, but this kinda takes the cake.





You guys are desperately trying to make up excuses and distract the issue away from a core fact which you just don't want to address. Do you honestly believe that 46% of the population of the US actually self-identifies as being Democrats? Because that would be astounding, don't you think? Step back from your own partisan idiocy and look at the damn number. It's not possible for 46% of the population to identify that way. Thus, any poll in which the sample polled gives that sort of response level must be skewed. And not just a little bit. A whole hell of a lot.

Sigh...

When one represents 46% and the other is 38%, there's more skew in one than the other.


When the question at hand is approval for the leader of a political party, one matters more than the other.


Are you deliberately being stupid? 46 is greater than 38. The correlation between political party and approval for a member/leader of that party is greater than that between ideology and said member/leader. What about those two factors is your brain simply unable to automatically recognize and comprehend? You can't just toss out random crap and think it means something.

Edited, May 25th 2011 5:27pm by gbaji

I have no clue how they managed it Joph. But surely we can both agree that 46% of the general voting population does not self-identify as Democrats. Right? At the end of the day, that's the only issue here.

Ok, look ya nut. Now I don't think you're actually going to accept any of this because anyone who has convinced himself that a liberal media conspiracy is covering for Obama as he stages (but not really stages!) situation room photos isn't going to let go of a simpler polling bias conspiracy. But that's fine.

First off, the poll itself has a margin of error of four points. This isn't even to say that it's absolute within that range and that there isn't some error. But then the accusation isn't in some statistical error, it's in the claim that this was intentionally skewed.

Secondly, Rasmussen has a known Republican bias. They don't admit to it or call themselves a Republican firm but their policy polling is generally well outside the norms of independent pollsters and their election polling (where you can actually hold them to a known result) gives them a bias of over 3 points in the Republican direction, about the worst of any major firm. I was just laughing at you for using Rasmussen as your baseline of what the numbers should look like. You probably didn't know this because, frankly, you don't know a lot of things and have little interest in learning when it's not going to support your point.

Third, as you completely misunderstood, ideological self-labeling is probably more significant than partisan labeling as people are more apt to shed partisan labels than change their ideology. You made the ridiculous assumption that this meant that there should be less Democrats because you made the (also ridiculous) assumption that this happened evenly across the board or something. The AP/Gfk results over time showed that the fluctuations mainly occurred between the Republicans and the "Can't/Won't Say" group. Self-identifying Democrats are relatively stable but when Republican identification is higher, the number is drawn from the third group and when it shrinks, the Democrats don't gain the numbers, the "Independents" do. So while you wail and shriek about the number of Republicans, the actual pool of people within that broader group is pretty much the same.

Why does that number change? Most likely because of people who identify with the Republican party but who don't want to consider themselves Republicans either because of some general dislike for the label or, more so, because they identify as Tea Partiers. Other polls have shown that the number of Tea Partiers who self-identify has Republicans is 55-70% or so. Discounting the fairly negligible amount of Tea Party Democrats, there's a fairly substantial pool of people who are Republican for all intents and purposes but who won't call themselves that. Democrats don't really have a second option for party identification.

The AP/Gfk poll is designed to push people into picking one group or the other. Tea Party Republicans have a better reason to refuse this push than "Independent" Democrats.

Actual approval across ideological lines isn't indicative of approval across partisan lines. The poll was taken to measure a post-bin Laden bounce when moderate conservatives & moderate Republicans would be more likely to approve of the president for at least a short time. You made the error earlier of trying to correlate the two.

So there ya go. There's certainly the chance that the AP/Gfk made some statistical error although almost certainly not as grave of one as you'd like to hope. There's really no reason to think you found evidence of some sort of intentional deck-stacking although reality is never as sexy as a good conspiracy theory so... well, have fun replying.

Joph. That's a wonderful explanation as to why the GOP polling sample appears so low. I'm not contesting that. Just as I'm not making hay about the Liberal numbers being low on the ideology scale thing. As you say, it's a label association thing.



But none of your long-winded post comes close to explaining how 46% self-reporting as Democrat isn't high. You talk about every single thing except the one number that matters. Strange, isn't it?


I'll ask the same "cut to the damn chase" question again:

Do you honesty believe that 46% of the general population identify themselves as Democrats? It's not a hard question, and we all know the answer.

Edited, May 25th 2011 5:50pm by gbaji

First off, alleging a skew in the Rasmussen poll doesn't counter an allegation of skew in the AP poll.

Secondly, that's pretty creative math though. The AP poll *also* has a "don't know/don't lean" category, which was 24%. Splitting those people towards a party in one, but not the other is kinda unfair, right?


Ignoring that bogus bit of math, the numbers just don't add up. 30% GOP is close to the 29% GOP in the AP poll. But 38% is nowhere freaking near 46%. Remember, I'm not arguing that they somehow failed to get a representative sample of Republicans, but that they have an abnormally high representation of Democrats.


Heck. Even your "Rasmussen is skewed by 3 points!" argument still backs me up, right? Apply an even 3% skew to the Rasmussen numbers (34% GOP, 33% Dem), and we're right about at the same numbers as the Reason poll, right? So we account for the claimed skew in Rasmussen's results, but we're still nowhere near accounting for a 46% Democrat identification in the AP poll.


I'm not arguing anything else Joph. I'm simply saying that the 46% Democrat identification number is too high and that this will/does skew the results in favor of Obama, as the leader of the Democratic Party. I'm not wrong about this.