HomeLatest ThreadsGreatest ThreadsForums & GroupsMy SubscriptionsMy Posts
DU Home » Latest Threads » Loki Liesmith » Journal
Page: « Prev 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Next »

Loki Liesmith

Profile Information

Member since: Thu May 26, 2016, 09:07 AM
Number of posts: 4,428

About Me

God of lies. Like math.

Journal Archives

An experiment in manipulating online media.


Can we make it trend on twitter? So that "people are talking" about it?

anyone game?
Posted by Loki Liesmith | Thu Sep 15, 2016, 10:37 AM (0 replies)

Stop with the "CNN didn't poll anyone under 50"

They do not report data for subgroups with margins of error above a certain threshold.
Posted by Loki Liesmith | Wed Sep 14, 2016, 08:46 PM (17 replies)

If I were Powell I'd endorse HRC immediately

He has a lot of shit talking to make up for.

He usually waits until later in the cycle to have more impact but hard to believe he could more impactful than right now.

Just my thoughts.
Posted by Loki Liesmith | Wed Sep 14, 2016, 07:31 PM (12 replies)

Yougov poll over weekend does not show pronounced "bad weekend effect"

H42 T40

About the same.


Still expect QPAC to be devastating for many here.
Posted by Loki Liesmith | Wed Sep 14, 2016, 12:50 PM (2 replies)

Average of the national polls this week (thus far)

So far looks a lot like last week:

NBC/SurveyMonkey NEW! Clinton +4
Morning Consult Clinton +1
ABC/Post Clinton +8

Average is 5, median is 2.5 for Clinton

However, two of these (ABC, Morning Whatever) are prior to the stupidfest this weekend.

UPI has been more pro-Trump than other polls, and has had him leading a few times already. Very sensitive to pro-Trump data.

My own model estimates that we should see a national spread between Clinton and
Trump of ~ 3.8 for Clinton.

Posted by Loki Liesmith | Tue Sep 13, 2016, 11:18 AM (9 replies)

Expect a CBS poll any day now

It's been a while (July 31? I think) since we've seen one.
Posted by Loki Liesmith | Mon Sep 12, 2016, 12:54 PM (2 replies)

There actually hasn't been much erosion in HRCs poll numbers

Take a look at this graphic:

(courtesy @willjordan, YouGov, on Twitter)

There has been an expansion of one of the tails of the distribution toward Trump, slightly. The body of the distribution is more or less where it's been for a long time. And the sampling density of polls was much higher near the conventions. If we had more frequent polling I am fairly sure we'd fill in the middle of that distribution more!
Posted by Loki Liesmith | Mon Sep 12, 2016, 12:03 PM (1 replies)

Some here really need to get a grip.

People vacillate from JOY to TERROR here so quickly, far more quickly than the fundamentals underlying this election can possibly change.

Do yourself a favor, whatever emotion you are currently feeling about this election, examine it. Are you letting the potential consequences of HRC losing overwhelm the actual object probability she does lose? Then put that emotion away, it doesn't affect anything but you.

Throw away whatever your gut feeling is. Go to http://elections.huffingtonpost.com/pollster and count up the states where HRC has a lead. Add up their electoral votes. Try it a few times. Randomly drop a few states from her corner Use electoral-vote.com if you need to see a graphical representation. How many times does she lose the election? Divide this by the number of runs you try.

That's the only number you need to worry about.
Posted by Loki Liesmith | Sun Sep 11, 2016, 01:47 PM (5 replies)

Election Model Update

Two models: one that only uses current polling, and one that takes the current poll as a starting point and and includes a random drift term...essentially projecting the current scenario forward to the election. See Original Model Writeup (below) for details.

Previous model writeups:

9/1/2016 update

8/16/2016 update

Update with drift included 8/6/2016

Original Model Writeup 8/3/2016

Model projections:

Probability of D win using current polling: 84.5%
average number of electoral votes: 312
median number of electoral votes: 314
most common electoral scenario: 317 electoral votes

Probability of D win projected forward to November election: 64.0%
average number of electoral votes: 288
median number of electoral votes: 289
most common electoral scenario: 289 electoral votes

Current Popular Vote Spread Estimate: +3.8% Clinton (estimated by multiplying statewide leads by turnout 2012 state turnout)

Analysis: Some minor tweaks to model, mostly in the code for grabbing data from HuffPost pollster. I missed an update last week, so I can't say any slide in HRC's poll numbers has been arrested. I can say that her numbers look about the same as two weeks ago, some slippage in the projected numbers, but that's due to changes in the polling variance over time. I use this variance to construct a daily random "drift" variable. In general, the variance in D leaning states' polling has gone up, the R leaning states show less variance. So it's more likely to randomly drift to a Trump win than a few weeks ago. The mean, median, and mode of electoral vote leads for Clinton haven't changed a lot though, which bodes well for continued stability.
Posted by Loki Liesmith | Sun Sep 11, 2016, 01:32 PM (5 replies)

WAPO/ABC poll: I don't think represents a change in the race

Report on poll here.

I think this is indicative of the ?Live Caller" vs. Internet or "Robo-caller" divide again.
All we have had for weeks are crappy internet polls and robo polls. Now we have a traditional survey to examine.

Let's just hope that Live caller interviews are indeed better, as I suspect they are.

Posted by Loki Liesmith | Sun Sep 11, 2016, 12:37 AM (5 replies)
Go to Page: « Prev 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Next »