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

[variant]

No, it doesn’t make them more likely to be wrong. That’s not what being an outlier means.

Sure it does.
Outlier - Wikipedia

In any statistical analysis you generally look to the mean of the data set rather than the outliers as they are more likely to be incorrect.

 

[Hammster]

Sure it does.
Outlier - Wikipedia

In any statistical analysis you generally look to the mean of the data set rather than the outliers as they are more likely to be incorrect.

While that’s true, you cannot consider different polls to be data sets unless everyone is polling the same way.

 

[variant]

While that’s true, you cannot consider different polls to be data sets unless everyone is polling the same way.

The data set is public opinion. Polling methodology leads to differn't results.

The question becomes why I should accept Rasmussen's polling methodology when it leads to significantly differn't results?

 

[Hammster]

The data set is public opinion. Polling methodology leads to differn't results.

It doesn’t work that way.

 

[variant]

It doesn’t work that way.

Sure it does. The object of the measurement is always the driver of the data set. If you can tell me why I should trust Rasmussen's measurements over the other polls then you might have a point.

 

[Jeffwhosoever]

This is the one of the best meta-pools out there: RealClearPolitics - Election Other - President Trump Job Approval

I don't see a single poll where Trump's approval is greater than 50%, at least if you go by the most recent polls.

 

[variant]

This is the one of the best meta-pools out there: RealClearPolitics - Election Other - President Trump Job Approval

I don't see a single poll where Trump's approval is greater than 50%, at least if you go by the most recent polls.

Yeah. So, why should we take Rasmussen 48% approval seriously when the average of the polls is 40%?

 

[Hammster]

Sure it does. The object of the measurement is always the driver of the data set. If you can tell me why I should trust Rasmussen's measurements over the other polls then you might have a point.

I’ve never made the argument that one should trust them over others.

But saying that the polls are a data set would be like comparing Pepsi bottles, Coke bottles and Dr. Pepper bottles. They may all be injection blow molded, but they’ll have different quality requirements. So saying that two of them are similar and one is way different will not make the one a bad bottle.

 

[Jeffwhosoever]

Look up central limit theorem if you want to comprehend polling statistics. That is why poll sampling is SAMPLING. Regardless of the actual statistic underlying the population, the CLT states that the sampling distribution is a normal distribution, better known as a gaussian distribution or "bell curve".

 

[Hammster]

Yeah. So, why should we take Rasmussen 48% approval seriously when the average of the polls is 40%?

How do you know it’s wrong. Different and wrong aren’t the same thing.

 

[Jeffwhosoever]

The benefit of Real Clear Politics is that in meta-analysis, you can see any bias in the data and obviously throw out poor samples. Trump's true approval appears to be more like 40% not 50%.

 

[variant]

I’ve never made the argument that one should trust them over others.

But saying that the polls are a data set would be like comparing Pepsi bottles, Coke bottles and Dr. Pepper bottles. They may all be injection blow molded, but they’ll have different quality requirements. So saying that two of them are similar and one is way different will not make the one a bad bottle.

Ah metaphors.

It depends. If the bottle's job is to hold a certain amount of liquid and one of them holds significantly more or less on average than any others, I am inclined to say that it indeed poorly designed.

 

[Hammster]

Ah metaphors.

It depends. If the bottle's job is to hold a certain amount of liquid and one of them holds significantly more or less on average than any others, I am inclined to say that it indeed poorly designed.

True. But you compare it to what the customer says is a good bottle, not other company’s bottles.

 

[variant]

How do you know it’s wrong. Different and wrong aren’t the same thing.

Because the other polls represent more measurements, most of which disagree. The poll in question is a measurement outlier and thus more likely to be incorrect.

Your premise is incorrect when you get to larger groups of estimations of any given value they tend to be more correct regardless of methodology or even expertise when they are significantly larger.

 

[variant]

True. But you compare it to what the customer says is a good bottle, not other company’s bottles.

Context is key. If the issue is about some task the bottle is actually designed to do then it can be measured with respect to other bottles designed for the same task.

 

[variant]

The benefit of Real Clear Politics is that in meta-analysis, you can see any bias in the data and obviously throw out poor samples. Trump's true approval appears to be more like 40% not 50%.

Rasmussen tells us something though, which is that Trump's popularity with the group that they tend to poll is usually higher.

 

[Jeffwhosoever]

Rasmussen said Romney was going to win as well, so perhaps they have some degree of GOP bias, but I also recall Rasmussen saying Hillary would win too.

 

[Hammster]

Because the other polls represent more measurements, most of which disagree. The poll in question is a measurement outlier and thus more likely to be incorrect.

Your premise is incorrect when you get to larger groups of estimations of any given value they tend to be more correct regardless of methodology or even expertise when they are significantly larger.

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

 

[Hammster]

Rasmussen said Romney was going to win as well, so perhaps they have some degree of GOP bias, but I also recall Rasmussen saying Hillary would win too.

They said Hillary would win the popular vote by 2%, which she did. They got Romney wrong, but were correct in ‘08 and ‘04.

 

[Jeffwhosoever]

They said Hillary would win the popular vote by 2%, which she did. They got Romney wrong, but were correct in ‘08 and ‘04.

Our Final 2016 Picks - Rasmussen Reports®