Trump's Approval = 50%: a good 5% above Obama's at same time in office

[variant]

They may represent more measurements. They may also ask questions in a way that skews the polls.

Most polls publish the questions. I couldn't find Rasmussen's questions on their website.

 

[Hammster]

Our Final 2016 Picks - Rasmussen Reports®

That’s the electoral college.

 

[Hammster]

Most polls publish the questions. I couldn't find Rasmussen's questions on their website.

Okay.

 

[Hammster]

Just to reiterate. I am NOT saying Rasmussen is the most accurate. I’ve just not seen anyone give any solid reason to reject them.

 

[variant]

Just to reiterate. I am NOT saying Rasmussen is the most accurate. I’ve just not seen anyone give any solid reason to reject them.

Well they don't publish their weighting methodology.

 

[Hammster]

Well they don't publish their weighting methodology.

Which isn’t a reason not to trust them.

 

[variant]

Which isn’t a reason not to trust them.

If you trust black boxes.

 

[Hammster]

If you trust black boxes.

My trust factor has nothing to do with accuracy.

 

[Jeffwhosoever]

That’s the electoral college.

Right. My point was that Rasmussen got even that part of the Hillary vs. Trump election wrong, not just the popular vote.

 

[variant]

My trust factor has nothing to do with accuracy.

Well when you ask a polling service how and why they do things and they say "it's proprietary" and then come up with usually differn't results that skew in a specific direction then you have a reason not to trust them.

For example PEW research has this to say about their sampling methodology:
Our survey methodology in detail

So we can actually examine what they are doing and their reasons for doing so to find out where their bias's are coming from.

Rasmussen:
Methodology - Rasmussen Reports®

Is much more vague about the whole thing.

 

[Hammster]

Right. My point was that Rasmussen got even that part of the Hillary vs. Trump election wrong, not just the popular vote.

They did get the popular vote right, just not where the votes came from.

White House Watch - Rasmussen Reports®

 

[Hammster]

Well when you ask a polling service how and why they do things and they say "it's proprietary" and then come up with usually differn't results that skew in a specific direction then you have a reason not to trust them.

For example PEW research has this to say about their sampling methodology:
Our survey methodology in detail

So we can actually examine what they are doing and their reasons for doing so to find out where their bias's are coming from.

Rasmussen:
Methodology - Rasmussen Reports®

Is much more vague about the whole thing.

And while that may weigh into whether or not you trust them, it has nothing to do with accuracy.

 

[variant]

And while that may weigh into whether or not you trust them, it has nothing to do with accuracy.

You were the one that brought up trust. And, trust has everything to do with accuracy, as you are trusting them to try to get you to the truth. Since they don't share well how they generate the results then you are left with only the results. Which, if you like your polls with a conservative slant, then approval rating for president is one of those for you with Rasmussen.

Accuracy is a measurement of predicted vs actual, so, since we don't measure any actual "approval" you can look up Rasmussen predictions to gauge their accuracy on events they end up predicting, otherwise this IS a matter of trust.

I don't actually understand your position other than being difficult for the sake of being difficult. If you aren't actually going to make a case for believing the outlier of Rasmussen then there really isn't much here to converse about.

I say it's an outlier because it is slanted that way usually, and should be taken with adequate skepticism.

 

[Hammster]

You were the one that brought up trust. And, trust has everything to do with accuracy, as you are trusting them to try to get you to the truth. Since they don't share well how they generate the results then you are left with only the results. Which, if you like your polls with a conservative slant, then approval rating for president is one of those for you with Rasmussen.

Accuracy is a measurement of predicted vs actual, so, since we don't measure any actual "approval" you can look up Rasmussen predictions to gauge their accuracy on events they end up predicting, otherwise this IS a matter of trust.

I don't actually understand your position other than being difficult for the sake of being difficult. If you aren't actually going to make a case for believing the outlier of Rasmussen then there really isn't much here to converse about.

I say it's an outlier because it is slanted that way usually, and should be taken with adequate skepticism.

Trust has nothing to do with accuracy. I can’t even imagine why one would think that.

And I’ve already stated my reason for posting here. It was pretty unambiguous.

 

[variant]

Trust has nothing to do with accuracy. I can’t even imagine why one would think that.

And I’ve already stated my reason for posting here. It was pretty unambiguous.

With regard to any endeavor to predict or represent reality trust and accuracy are one in the same.

Your magical skepticism of more well founded skepticism than your own is noted but frankly uninteresting.

 

[Hammster]

With regard to any endeavor to predict or represent reality trust and accuracy are one in the same.

Your magical skepticism of more well founded skepticism than your own is noted but frankly uninteresting.

Something is either accurate or it’s not. Whether someone trusts it is irrelevant. Otherwise, you’d have to admit that Rasmussen is accurate since a lot of people trust them.

 

[KCfromNC]

Well when you ask a polling service how and why they do things and they say "it's proprietary" and then come up with usually differn't results that skew in a specific direction then you have a reason not to trust them.

Plus the questionable results in the cases we can measure against actual elections.

 

[Hammster]

Plus the questionable results in the cases we can measure against actual elections.

If that’s a marker, which polls got it right in ‘16?

 

[bhsmte]

If that’s a marker, which polls got it right in ‘16?

I don't think anyone got it right, did they?

 

[Jeffwhosoever]

Rasmussen is just another data source, no different than any other actual SAMPLING of the true population, because you can't spend the money to call every person in the United States. Any sampling has limitations as revealed by Rasmussen's projection that Hillary would win over 300 electors in the vote, but they have been right on some things, as has any other data source. But always be skeptical of sampled data, because even without bias, it is subject to many sources of error. In the case of Trump vs. Clinton, I don't think anyone projected Trump to win in the electoral system. However, ultimately he won about what Hillary was projected to win, due to a brilliant campaign sweep of the "rust belt" states in the final stages that swayed just enough voters, plus winning Florida. I say that trying to pick a close election like that is like trying to get a 7 day forecast of the weather 100% correct - it just is blind luck in the end, because no amount of statistics and sampling theory and process can account for all the variables. When they say the data is valid to +/- 4%, that means they are just as likely off by +4% as off by -4% and the statistics cannot tell you which is correct. Trying to infer how 100,000,000 or more voters will choose on election day from surveying 6,000 potential voters a few days in advance obviously has some inherent inaccuracies. Hence, my suggestion is to always look at meta-surveys, like RealClearPolitics, and get a feel for both the dispersion (variance or standard deviation) as well as the trend ( the slope of the projected mean as it progresses in time). But the more data sources, the better, because things like poor sampling choices, bias, etc. get filtered out in the aggregate and you can get more accurate results because it is unlikely that the same people get asked by all the different polling services like Rasmussen, and as the number of samples n approaches the entire population size, the sample mean approaches the true population mean. That is, the more people that get included in the poll of polls, the more it represents what the entire population is thinking. So ultimately, it isn't a good practice to base all your inferences on any one source when so many others are available, and hence my recommendation to use the "poll of polls" rather than a single poll whether it be Rasmussen, Gallup, WSJ, or whatever.