Is teh LA Times/USC Tracking Poll,,,,,,,,

[Cryptoad]

gaming a Trump bias in its results for no other reason than to keep the Poll averages for Hillary down. I know there are outliers,,,,, but this poll is out in Right field all the time. which is at least twice a week. Something stinks in Denmark!

[SCliberal91294]

1. 7 day tracking poll

We will see gains in a few days.

[Loki Liesmith]

2. It's probably not a malicious effect

If you read the "polls" methodology, they made a serious mistake at the outset.

First off, recall that they are sampling the same panel over and over again. So any bias there at the beginning stays through the end.

Secondly, they asked their respondents who they voted for in 2012 and tried to match their respondent pool Obama's win numbers. This was a huge mistake. There is a well known "winner halo effect" by which a certain percentage of respondents will claim to have voted for the winner in past elections when they actually voted the other way. I have heard numbers for this effect ranging between 5 and 10% of respondents.

Lo and behold, the LA Times tracking poll appears to have a Trump +6 bias built in. Those are likely Romney voters who have had a convenient lapse in memory (because everyone wants to be a winner).

So no, I don't think they are trying to manipulate results. They just made a mistake at the outset and there is nothing to be doen about it now. I'm sure they even realize this at this point.

[still_one]

3. Realizing that their sampling and model is flawed, and not adjusting it for those errors is an

intentional manipulation in my view

[Loki Liesmith]

4. It was an experimental methodology

They are going to evaluate the experiment at the end in terms of its predictive quality.

I think that changing the methodology mid stream would actually be dishonest.

Also if this is grant funded work, they may NOT be allowed to completely change the methodology.

All the data is there for them at the end to play with. They can write about it in the paper that they publish.

[Loki Liesmith]

5. It's also good the poll averages

to have some consistent outliers. Let's us figure out which methodology is most robust to bad data.

[Cryptoad]

6. you trust those Republican SonOf Beachs far more that I do!

[MFM008]

7. YES

Yyyyyyyeeeeeeeesssssssssss.
She has been ahead by one point I think a couple of times.
The rest have all been for the maggot.