The General Election

[Danny Bateman]

Trump University, part Deux: The Trump Institute.

 

http://nytimes.com/2016/06/30/us/politics/donald-trump-institute-plagiarism.html?smid=tw-nytimes&smtyp=cur&_r=0&referer=https://t.co/ODa2rLBFQy

 

https://twitter.com/nytimes/status/748311977479536640

[HuskerNation1]

 

 

 

It looks like Trump and Clinton are back to being deadlocked.

 

http://www.nytimes.com/2016/06/30/us/politics/hillary-clinton-and-donald-trump-are-deadlocked-poll-shows.html?_r=0

 

http://www.qu.edu/news-and-events/quinnipiac-university-poll/national/release-detail?ReleaseID=2363

 

As I stated before, I think this thing is going to go back and forth. Most voters don't really focus on the candidates and their positions on issues until labor day or after, so I think at that point in time some of these national and state numbers will be more meaningful. What is most intriguing about this poll is that Trump is performing quite a bit better among Hispanics than Romney or McCain did.

If you look at individual polls, you are probably correct.

However, I prefer to look at websites that combine polls together.

 

It's similar to me liking 247 composite ratings of recruits better than any individual rating site.

 

 

Oh I agree I typically look at the RCP aggregate, but I also look at polling trends. If we see another poll come out in the next few days that shows the race at 2-5 points instead of 9-11 (like some were showing last week), I see that as a trend change. One thing that RCP does not do, however, is remove outliers where the numbers are so far off base they skew the entire average heavily.

 

 

Outliers are quite normal, to be honest. It's not necessarily good polling to remove them, nor good statistics. In any stats work I did back in school, they taught us to try to explain how outliers happened rather than exclude them.

 

n short, though, some variability is good. Nate Silver and Harry Enten com 538 have been hammering this point on Twitter.

 

 

Well I think it depends on your stats class. I was a stats tutor for a while and helped some of the juniors with their Econ research projects, and we often focused on data quality to ensure certain data samples were not skewing the overall output. Everyone does it different though, and it depends on what you are trying to measure. While Nate thinks pretty highly of himself and was on point in 2012, he has had some misses including the 2016 GOP nomination which resulted in him explaining why he was so wrong. I think if we had a traditional GOP nominee like Jeb Bush, Nate and most other pundits and pollsters would be more on point, but the Trump movement is so atypical that I don't think any poll will accurately project his turnout in November.

 

http://fivethirtyeight.com/features/how-i-acted-like-a-pundit-and-screwed-up-on-donald-trump/

[Red Five]

It looks like Trump and Clinton are back to being deadlocked.

 

 

Not so fast my friend...

 

Clinton +6 in Fox News poll tonight. Basically the average of the past couple weeks of polls.

 

​http://www.foxnews.com/politics/2016/06/29/fox-news-poll-clinton-up-by-6-points-89-percent-say-hot-headed-describes-trump.html

[HuskerNation1]

 

It looks like Trump and Clinton are back to being deadlocked.

 

 

Not so fast my friend...

 

Clinton +6 in Fox News poll tonight. Basically the average of the past couple weeks of polls.

 

​http://www.foxnews.com/politics/2016/06/29/fox-news-poll-clinton-up-by-6-points-89-percent-say-hot-headed-describes-trump.html

 

If you quoted the rest of what I stated, its going to go back and forth, and I believe Clinton has a lead of 4-6 currently, but it's early and many voters do not concentrate until after Labor Day. I fully suspect Hillary to keep a small lead for some time.

[Moiraine]

 

 

 

Oh I agree I typically look at the RCP aggregate, but I also look at polling trends. If we see another poll come out in the next few days that shows the race at 2-5 points instead of 9-11 (like some were showing last week), I see that as a trend change. One thing that RCP does not do, however, is remove outliers where the numbers are so far off base they skew the entire average heavily.

 

 

Outliers are quite normal, to be honest. It's not necessarily good polling to remove them, nor good statistics. In any stats work I did back in school, they taught us to try to explain how outliers happened rather than exclude them.

 

n short, though, some variability is good. Nate Silver and Harry Enten com 538 have been hammering this point on Twitter.

 

 

Well I think it depends on your stats class. I was a stats tutor for a while and helped some of the juniors with their Econ research projects, and we often focused on data quality to ensure certain data samples were not skewing the overall output.

 

 

Basically... 95% of identically created confidence intervals will contain the true margin of victory, but if the standard error is high we could have a lot of different looking values within those confidence intervals. I haven't really looked at margins-of-error in the polls and have no idea about the variation among the different polls themselves.

 

The biggest issue is the sample has to be random for the above to be 100% true, and most samples aren't. Each polling company likely has a different set of people on their lists, then we have some who are asking RVs instead of LVs, and there is a lot of research going on right now about people who don't answer their phone. Should they really be dropped from the survey? Does that make it no longer random? Is there something about those people that makes them not answer the phone? We could be missing something important about those people that is an indicator of how they might vote. Then there is likely still the problem of reaching people who only use cell phones. You're missing out on a good chunk of the population that way.

[mrandyk]

 

I am convinced more now than ever that Trump's only purpose for running was to help Hillary get elected. He has done nothing to exhibit he really wants the office.

I've been saying that since day one. I've never taken him seriously as a candidate.

 

I am not convinced that his purpose was to help Hillary, I have felt that promoting his name has been top priority. Definitely not a guy that anticipated winning the nomination at any rate.

[HuskerNation1]

One rule of thumb I've always had.

 

If someone has to constantly tell me how smart or how great they are....well....I'm not buying it.

 

Or focus on how bad someone is (as Obama pointed out his entire first term about Bush), I'm not buying that either. There is no need for any politician to put others down to make themselves appear better.

[HuskerNation1]

I think Trump's firing of Lewandoswki is starting to pay off for him. Sure, he will continue to make bone-headed statements, but he appears to be more focused and disciplined and on message, so much so that the Huffington Post could not help but refrain from blasting his trade speech today.

 

http://www.huffingtonpost.com/ian-fletcher/trump-delivers-major-trad_b_10721588.html?yptr=yahoo

 

I have to say if he can focus on how bad this economic recovery has been (especially for blue-collar workers) as well as how much worse off we are regarding terrorism, he is going to beat Hillary in November. I just don't know if he's got the discipline to hammer these 2 central messages over and over.

[HuskerShark]

I think Trump's firing of Lewandoswki is starting to pay off for him. Sure, he will continue to make bone-headed statements, but he appears to be more focused and disciplined and on message, so much so that the Huffington Post could not help but refrain from blasting his trade speech today.

 

http://www.huffingtonpost.com/ian-fletcher/trump-delivers-major-trad_b_10721588.html?yptr=yahoo

 

I have to say if he can focus on how bad this economic recovery has been (especially for blue-collar workers) as well as how much worse off we are regarding terrorism, he is going to beat Hillary in November. I just don't know if he's got the discipline to hammer these 2 central messages over and over.

I think he's very perceptive as to what works and what doesn't and follows whichever trend that gives him a favorable response.

 

That's not saying that he changes his stances, as many seem to think. He simply changes the way he presents the information to get the best reaction, like any intelligent sales person would do.

[Danny Bateman]

 

I think Trump's firing of Lewandoswki is starting to pay off for him. Sure, he will continue to make bone-headed statements, but he appears to be more focused and disciplined and on message, so much so that the Huffington Post could not help but refrain from blasting his trade speech today.

 

http://www.huffingtonpost.com/ian-fletcher/trump-delivers-major-trad_b_10721588.html?yptr=yahoo

 

I have to say if he can focus on how bad this economic recovery has been (especially for blue-collar workers) as well as how much worse off we are regarding terrorism, he is going to beat Hillary in November. I just don't know if he's got the discipline to hammer these 2 central messages over and over.

I think he's very perceptive as to what works and what doesn't and follows whichever trend that gives him a favorable response.

 

That's not saying that he changes his stances, as many seem to think. He simply changes the way he presents the information to get the best reaction, like any intelligent sales person would do.

 

 

On the other hand...

 

Was he in intelligent sales person mode when he was assailing a federal judge as professionally incompetent due to his race? Repeated boldfaced racism does not seem to be bering perceptive to what works to me...

[Danny Bateman]

 

 

 

 

It looks like Trump and Clinton are back to being deadlocked.

 

http://www.nytimes.com/2016/06/30/us/politics/hillary-clinton-and-donald-trump-are-deadlocked-poll-shows.html?_r=0

 

http://www.qu.edu/news-and-events/quinnipiac-university-poll/national/release-detail?ReleaseID=2363

 

As I stated before, I think this thing is going to go back and forth. Most voters don't really focus on the candidates and their positions on issues until labor day or after, so I think at that point in time some of these national and state numbers will be more meaningful. What is most intriguing about this poll is that Trump is performing quite a bit better among Hispanics than Romney or McCain did.

If you look at individual polls, you are probably correct.

However, I prefer to look at websites that combine polls together.

 

It's similar to me liking 247 composite ratings of recruits better than any individual rating site.

 

 

Oh I agree I typically look at the RCP aggregate, but I also look at polling trends. If we see another poll come out in the next few days that shows the race at 2-5 points instead of 9-11 (like some were showing last week), I see that as a trend change. One thing that RCP does not do, however, is remove outliers where the numbers are so far off base they skew the entire average heavily.

 

 

Outliers are quite normal, to be honest. It's not necessarily good polling to remove them, nor good statistics. In any stats work I did back in school, they taught us to try to explain how outliers happened rather than exclude them.

 

n short, though, some variability is good. Nate Silver and Harry Enten com 538 have been hammering this point on Twitter.

 

 

Well I think it depends on your stats class. I was a stats tutor for a while and helped some of the juniors with their Econ research projects, and we often focused on data quality to ensure certain data samples were not skewing the overall output. Everyone does it different though, and it depends on what you are trying to measure. While Nate thinks pretty highly of himself and was on point in 2012, he has had some misses including the 2016 GOP nomination which resulted in him explaining why he was so wrong. I think if we had a traditional GOP nominee like Jeb Bush, Nate and most other pundits and pollsters would be more on point, but the Trump movement is so atypical that I don't think any poll will accurately project his turnout in November.

 

http://fivethirtyeight.com/features/how-i-acted-like-a-pundit-and-screwed-up-on-donald-trump/

 

 

You're right about Trump. He's been so starkly anomalous that I'd frankly be shocked if anyone (or any statistical model) had him winning the way he did based on anything other than a hunch. He's the exception to the rule.

 

It was probably just my experience with stats. I didn't venture very deep into that pool during my undergrad. I just remember distinctly that when we had outliers in my capstone project as a senior, they wouldn't let us discard obvious outliers in our data set. The fact we had to try to reason their origin may have been because we were publishing our work. I'm not sure how they treat outliers in heavier duty stats classes... thanks for the insight!

 

Silver isn't perfect, and has been worse, but a lot of that is Trump. He's still the best in the biz at what he does.

[HuskerShark]

 

 

 

I think Trump's firing of Lewandoswki is starting to pay off for him. Sure, he will continue to make bone-headed statements, but he appears to be more focused and disciplined and on message, so much so that the Huffington Post could not help but refrain from blasting his trade speech today.

 

http://www.huffingtonpost.com/ian-fletcher/trump-delivers-major-trad_b_10721588.html?yptr=yahoo

 

I have to say if he can focus on how bad this economic recovery has been (especially for blue-collar workers) as well as how much worse off we are regarding terrorism, he is going to beat Hillary in November. I just don't know if he's got the discipline to hammer these 2 central messages over and over.

I think he's very perceptive as to what works and what doesn't and follows whichever trend that gives him a favorable response.

 

That's not saying that he changes his stances, as many seem to think. He simply changes the way he presents the information to get the best reaction, like any intelligent sales person would do.

On the other hand...

 

Was he in intelligent sales person mode when he was assailing a federal judge as professionally incompetent due to his race? Repeated boldfaced racism does not seem to be bering perceptive to what works to me...

He definitely didn't word it right, but he wasn't wrong on that issue, and like everything he does, it was blown WAY out of proportion.

[NM11046]

 

 

 

I think Trump's firing of Lewandoswki is starting to pay off for him. Sure, he will continue to make bone-headed statements, but he appears to be more focused and disciplined and on message, so much so that the Huffington Post could not help but refrain from blasting his trade speech today.

 

http://www.huffingtonpost.com/ian-fletcher/trump-delivers-major-trad_b_10721588.html?yptr=yahoo

 

I have to say if he can focus on how bad this economic recovery has been (especially for blue-collar workers) as well as how much worse off we are regarding terrorism, he is going to beat Hillary in November. I just don't know if he's got the discipline to hammer these 2 central messages over and over.

I think he's very perceptive as to what works and what doesn't and follows whichever trend that gives him a favorable response.

 

That's not saying that he changes his stances, as many seem to think. He simply changes the way he presents the information to get the best reaction, like any intelligent sales person would do.

On the other hand...

 

Was he in intelligent sales person mode when he was assailing a federal judge as professionally incompetent due to his race? Repeated boldfaced racism does not seem to be bering perceptive to what works to me...

He definitely didn't word it right, but he wasn't wrong on that issue, and like everything he does, it was blown WAY out of proportion.

 

Shark, is the weather nice in the bubble?

[Red Five]

 

 

It looks like Trump and Clinton are back to being deadlocked.

 

 

Not so fast my friend...

 

Clinton +6 in Fox News poll tonight. Basically the average of the past couple weeks of polls.

 

​http://www.foxnews.com/politics/2016/06/29/fox-news-poll-clinton-up-by-6-points-89-percent-say-hot-headed-describes-trump.html

 

If you quoted the rest of what I stated, its going to go back and forth, and I believe Clinton has a lead of 4-6 currently, but it's early and many voters do not concentrate until after Labor Day. I fully suspect Hillary to keep a small lead for some time.

 

And you also said they were "deadlocked" when one poll came out +2

 

:shrug:

[Moiraine]

 

 

 

 

 

It looks like Trump and Clinton are back to being deadlocked.

 

http://www.nytimes.com/2016/06/30/us/politics/hillary-clinton-and-donald-trump-are-deadlocked-poll-shows.html?_r=0

 

http://www.qu.edu/news-and-events/quinnipiac-university-poll/national/release-detail?ReleaseID=2363

 

As I stated before, I think this thing is going to go back and forth. Most voters don't really focus on the candidates and their positions on issues until labor day or after, so I think at that point in time some of these national and state numbers will be more meaningful. What is most intriguing about this poll is that Trump is performing quite a bit better among Hispanics than Romney or McCain did.

If you look at individual polls, you are probably correct.

However, I prefer to look at websites that combine polls together.

 

It's similar to me liking 247 composite ratings of recruits better than any individual rating site.

 

 

Oh I agree I typically look at the RCP aggregate, but I also look at polling trends. If we see another poll come out in the next few days that shows the race at 2-5 points instead of 9-11 (like some were showing last week), I see that as a trend change. One thing that RCP does not do, however, is remove outliers where the numbers are so far off base they skew the entire average heavily.

 

 

Outliers are quite normal, to be honest. It's not necessarily good polling to remove them, nor good statistics. In any stats work I did back in school, they taught us to try to explain how outliers happened rather than exclude them.

 

n short, though, some variability is good. Nate Silver and Harry Enten com 538 have been hammering this point on Twitter.

 

 

Well I think it depends on your stats class. I was a stats tutor for a while and helped some of the juniors with their Econ research projects, and we often focused on data quality to ensure certain data samples were not skewing the overall output. Everyone does it different though, and it depends on what you are trying to measure. While Nate thinks pretty highly of himself and was on point in 2012, he has had some misses including the 2016 GOP nomination which resulted in him explaining why he was so wrong. I think if we had a traditional GOP nominee like Jeb Bush, Nate and most other pundits and pollsters would be more on point, but the Trump movement is so atypical that I don't think any poll will accurately project his turnout in November.

 

http://fivethirtyeight.com/features/how-i-acted-like-a-pundit-and-screwed-up-on-donald-trump/

 

 

You're right about Trump. He's been so starkly anomalous that I'd frankly be shocked if anyone (or any statistical model) had him winning the way he did based on anything other than a hunch. He's the exception to the rule.

 

It was probably just my experience with stats. I didn't venture very deep into that pool during my undergrad. I just remember distinctly that when we had outliers in my capstone project as a senior, they wouldn't let us discard obvious outliers in our data set. The fact we had to try to reason their origin may have been because we were publishing our work. I'm not sure how they treat outliers in heavier duty stats classes... thanks for the insight!

 

Silver isn't perfect, and has been worse, but a lot of that is Trump. He's still the best in the biz at what he does.

 

 

You don't discard them. Not if you want your results to be considered valid or want to get published and then not have a scandal later. Unless maybe you know for a fact that it's an outlier because the data collector screwed up horribly on that particular observation.

 

I seem to recall having one example where we were looking at a certain type of fish and its size. I can't remember what the predictor variables were, but we had one observation that was extremely far from the others and it turned out that particular observation was a much larger fish (different kind than was being studied) that got caught in the net. It could be kicked out of the data because it wasn't the type of fish that was being studied.