Could Someone Please Explain the Polling Disparaty at Reuters/Ipsos and RCP?

[louis c]

I follow the polling in this Presidential race with near obsession. That's because I love politics and I know this election truly is the most important in my lifetime.

Having said that, why do I read a Reuters/Ipsos national poll that has Hillary up by 8 points head to head with Trump and 7 points with Stein and Johnson in it dated Aug. 18?

Link:
http://www.reuters.com/article/us-usa-election-poll-idUSKCN10U1YI

And the Real Clear Politics Web Site that has a national Reuters/Ipsos poll that has Hillary 5 points over Trump head to head and 4 points with Johnson and Stein

Link:
http://www.realclearpolitics.com/epolls/latest_polls/president/

[Bernardo de La Paz]

1. RCP version of Reuters is old. . . nt

[louis c]

2. It's Thursday's

The newer version is Friday? That doesn't make sense, it's not a tracking poll, is it?

[Bernardo de La Paz]

3. RCP version of Reuters one back. It has a date the 17th. The Reuters link is released Friday 20th.

The older version was sampled Aug. 13-17.

The newer version was sample 14-18.

[major debacle]

6. The two results are from the same poll.

RCP provides a link to a PDF of the poll. Both results are present in the PDF:

http://www.realclearpolitics.com/docs/2016_Reuters_Tracking_-_Core_Political_8.17_.16_.pdf

All four results are in the PDF:

C vs T - LV
C vs T - RV
C vs T vs J vs S - LV
C vs T vs J vs S - RV

[major debacle]

7. Date discrepancy

The 8/14-18 date referred to in the Reuters article on the poll overlaps the 8/13-17 date on the poll itself.

I was not able to find a Reuters/Ipsos poll that is dated 8/14-18. Could the 8/14-18 date in the article be referring to the 8/13-17 poll?

In any case, the difference in the RCP version of the poll and the Reuters article version of the poll is between Likely Voters (RCP) and Registered Voters (Reuters article).

[Bernardo de La Paz]

8. You identified the Likely vs Reg cause of issue. Yeah, seems sloppy reporting on the dates. . nt

[major debacle]

9. "seems sloppy reporting on the dates."

I'm shocked...

[major debacle]

4. Not sure, working from memory

... but I believe the difference in the two versions can be attributed to 'registered voters' versus 'likely voters'.

After looking it up, my memory is correct (whew). The higher Clinton score is among registered voters, the lower among likely voters.

[Bernardo de La Paz]

5. Yes, see page 10 of the RCP link. Both are there. . . . nt

[louis c]

10. Now I understand, Thanks

I remember now that RCP works with likely voters rather than registered voters when they have a choice. But the articles sure are confusing.

You would think that Reuters would differentiate between the two in it's summary.

[Bernardo de La Paz]

11. Took me a while to sort it out (with help of Major) because of the misdirection from bad reporting.