Do not trust Real Clear Politics polling average

[jimmy the one]

Rasmussen & Trafalgar are both crap polls (Computer Responsive Automated Polling), as well as another sense of crap. Automated telephone polls, no human interaction. They both are republican pollsters and pad to the right.
When trump was pres, rasmussen padded his approval numbers by 6 to 8 pts above the rest of the field average. While Biden's been president, rasmussen has counter padded his approval by 8 to 10 pts lower. This is corroborated by nate silver's 538 poll website, which includes rasmussen in its presidential approval poll average, but anti-biases rasmussen, currently by 8 pts.
Real Clear Politics (RCP) also has a presidential approval poll average, but does not ant-bias pollster results, it lists all polls at face value, including rasmussen which skews the poll to the right consistently.

Here is the mar 04, 2022 rasmussen poll from '538' pollsters: :

Rasmussen 1,500LV 42% .. 55% /// adjusted:-44% .. 49%

Note rasmussen's face value 42% approval for Biden, with 55% dissapproval; so 55 - 42 = - 13 net difference. Note the adjusted value which '538' assigns as 44% approval to 49% disapproval, for a -5 net difference. So rather than -13 net disapproval, 538 adjusts rasmussen to only -5 disapproval, which is what is included in their daily poll average. On the RCP polling average, the -13 net diff is what they include for rasmussen. Biden bashing rasmussen padded 8 pts to biden's disapproval, according to 538's historical analysis of rasmussen (note: even tho other reputable polls currently have bidens' net approval ratings worse than rasmussens, is because rasmussen strategically pads to try to appear reputable when other pollsters are showing poor results for dems). https://projects.fivethirtyeight.com/biden-approval-rating/?ex_cid=rrpromo

Here's the mar 04 2022 marist poll which showed an 8 pt biden bump after sotu address:

Marist: 47% appr ... 50% disapproval /// adjusted 48% .. 50%

Marist has a -3 net disapproval for biden, which 538 anti-biased by 1 (marist leaned 1 to R, actually pretty fair pollster) to 48 to 50 for an adjusted net diff of -2.
The adjusted net difference is what 538 includes in it's presidential poll averages. RCP includes face value, which allows crappy pollsters like rasmussen and trafalgar and harris (R) to cheat and skew RCP average to the right, by about 1 or 2 pts better for republican and worse for democrats in the poll average (since there are usually a dozen or so polls in the polling average, it doesn't reflect as a larger comparative difference thanks for that anyway)
-----------------------
Rasmussen also is included in EVERY SINGLE PRESIDENTIAL APPROVAL POLL done by 538 and RCP. That is because rasmussen is a tracking poll and posts a (padded) presidential poll result every working day as it averages a rolling 3 day track.
Most reputable pplls post a presidential approval poll only once a month, like pew, gallup, quinn, nbc, cbs, abc and yes fox in this case. However, the daily RCP and 538 polling average only includes polls within a drop-back period of approx 18 days, wherein if the poll is older than 18 days it is no longer included in the polling average. Thus more reputable polls are excluded after 18 days and are not included almost half the time, while the padding rasmussen is included EVERY SINGLE POLL.
Since 538 'adjusts', it compensates, but Real Clear Politics (RCP) does not adjust or anti-bias and allows it's poll to be skewed by rasmussen. That is because real clear politics is better pronounced real queer politics.

Proof: Mar 4 presidential polling avg from '538': 52.4 disappr, 42.6 appr = net 9.8 disapproal
........ Mar 4 pres polling avg Real Clear Politics: 53.5 disappr, 41.6 appr = net 11.9 disapproval

That, ladies and gentlemen, is the rasmussen padding effect.
(both poll averages often are closer & infrequently RCP will even do better for dems, since they might be using different pollsters for that time period, but when the same pollsters are used in the avg, RCP sucks for dems)

Having a bias of 1 or 2 pts is pretty normal. But 8 pts is abnormal. Trafalgar was 5 pts R, there is one pollster 4 D, I forget which.

[acantharchus]

1. Silver's adjusted avg is credible; Rasmussen, garbage. Thanks. Those in the dark need to know. NT

[secondwind]

2. K&R for visibility

[Ouroborosnek]

3. Good info thanks!

Every time I see a poll, I keep wishing I could participate in one of these.

[FBaggins]

4. Not liking results that a pollster reports is not the same thing as that firm being bad at polling

Neither is being a pollster who mainly works with one party or the other.

You're misunderstanding the "bias" adjustment. It doesn't mean that a pollster is "off" in a given direction (on average) - it compares their common results with that of other pollsters.

The problem with that assumption is that both of these pollsters (particularly Trafalgar) have been much closer to the actual results for a few cycles now. They didn't align with the other pollsters... but it was the other pollsters who were getting it wrong.

Ignoring them because it makes us feel better could be a perilous mistake (it certainly was in Virginia last year and in more than one critical race in 2020)

[jimmy the one]

6. rasmussen strategically pads to the right

Baggins: You're misunderstanding the "bias" adjustment. It doesn't mean that a pollster is "off" in a given direction (on average) - it compares their common results with that of other pollsters.

538 does not call it anti-biasing. I personally refer to it as 'anti-biasing' since that's what it appears to be. Fox is slightly 'anti-biased' by 538 as being pro R, msnbc as pro D, etc... generally. Since the approval polls are adjusted over time perforce to a particular party, it seems a fair way to put it. A 1 or 2 pt 'anti-bias' should not infer the poll is invalid or truly biased R or L.

The problem with that assumption is that both of these pollsters (particularly Trafalgar) have been much closer to the actual results for a few cycles now. They didn't align with the other pollsters... but it was the other pollsters who were getting it wrong.

What do you mean by actual results? for an approval poll? election results perhaps, but that is not what I am talking about in this thread. Rasmussen amongst others strategically pads with intent to coyly deceive. While rasmussen padded trump 2016 - 2020 they might've counter padded a down ballot R race to make it seem they were fair and balanced. While this also would indeed detract from their accuracy, it wouldn't too much if rass was coy & careful, I think I saw a 78% accuracy rating from 538. Rasmussen had GWBush winning over al gore by 9 pts, the worst of the major pollsters. It vowed to do better and in subsequent years it was pretty close in accuracy. Which only bolsters my belief that Rass strategically pads, especially in love with trump and against any Dem president or candidate.

Trafalgar Group head pollster Robert Cahaly told Fox News that he thinks President Trump will win the November 3 election despite most polls showing Joe Biden ahead. "What we’ve noticed is that these polls are predominantly missing the hidden Trump vote. There is a clear feeling among conservatives and people that are for the president that they’re not interested in sharing their opinions readily,,". https://www.realclearpolitics.com/video/2020/10/21/trafalgar_group_head_pollster_predicts_trump_will_win_polls_missing_hidden_trump_vote.html

Rasmussen Reports is an American conservative polling company founded in 2003

Rasmussen polls are often viewed as outliers due to their favorable Trump approval ratings.

For the 2020 presidential election, Rasmussen final survey of likely U.S. voters showed Biden with a 1% lead over Trump, stating that ".. a near tie." Biden won by 4.5 percentage points {9.6% more popular votes}. After Trump lost the election, he made false and disproven claims of fraud and attempted to overturn the results, with Rasmussen itself suggesting that Vice President Mike Pence should overturn the election results

2018, Rasmussen Reports predicted that Republicans would win the generic ballot by 1 point while the actual election results had Democrats winning by nearly 9 points.

For surveys such as its daily Presidential Tracking Poll, Rasmussen's automated technology calls randomly selected phone numbers.. To reach those who have abandoned landlines, an online survey tool interviews randomly selected participants from a demographically diverse panel. These types of polls are believed to be less accurate and biased.

Rasmussen is not a member of the National Council on Public Polls or a supporter of the American Association for Public Opinion Research's Transparency Initiative

The accuracy of Rasmussen's polling has varied considerably in recent U.S. presidential elections (2000–2016)... FiveThirtyEight gave the firm an overall rating of "B," reporting it had a 1.5-point bias in favor of the Republican Party.
https://en.wikipedia.org/wiki/Rasmussen_Reports#:~:text=2020-,For%20the%202020%20United%20States%20presidential%20election%2C%20Rasmussen%20Reports'%20final,the%20election%20by%204.5%20percentage

The 1.5 point R bias above understates rasmussen's bias for presidential approval ratings. When 538 adjusts rass' approval rating by 8 pts, sounds like anti biasing to me.

[DarthDem]

5. K&R for Viz

Thanks and I agree.

[jimmy the one]

7. Criticism of right wing Rasmussen, C.R.A.P. poll extraordinaire

Criticism of Rasmussen Reports

FiveThirtyEight In 2010, Silver wrote an article titled "Is Rasmussen Reports biased?" in which he mostly defended Rasmussen from allegations of bias. However, later in the year, Rasmussen's polling results diverged notably from those of other mainstream pollsters, which Silver labeled a "house effect." He went on to explore other factors that may have explained the effect, such as the use of a likely voter model, and claimed that Rasmussen conducted its polls in a way that excluded the majority of the population from answering.
After the 2010 midterm elections, Silver concluded that Rasmussen's polls were the least accurate of the major pollsters in 2010, having an average error of 5.8 points and a pro-Republican bias of 3.9 points a ccording to Silver's model.[83]

Bias in questions Jonathan Chait of the New Republic said that Rasmussen is perceived in the "conservative world" as "the gold standard" and suggested the polling company asks the questions specifically to show public support for the conservative position.

The Center for Public Integrity listed "Scott Rasmussen Inc" as a paid consultant for the 2004 George W. Bush campaign. The Washington Post reported that the 2004 Bush re-election campaign had used a feature on the Rasmussen Reports website that allowed customers to program their own polls, and that Rasmussen asserted that he had not written any of the questions nor assisted Republicans.

In 2009 Time magazine described Rasmussen Reports as a "conservative-leaning polling group." John Zogby said in 2010 that Scott Rasmussen had a "conservative constituency." In 2012 The Washington Post called Rasmussen a "polarizing pollster."

Rasmussen has received criticism over the wording in its polls. Asking a polling question with different wording can affect the results of the poll; the commentators in question allege that the questions Rasmussen ask in polls are skewed in order to favor a specific response. For instance, when Rasmussen polled whether Republican voters thought Rush Limbaugh was the leader of their party, the specific question they asked was: "Agree or Disagree: 'Rush Limbaugh is the leader of the Republican Party—he says jump and they say how high.'"

Talking Points Memo has questioned the methodology of Rasmussen's Presidential Approval Index, which takes into account only those who "strongly" approve or disapprove of the President's job performance. TPM noted that this inherently skews negative, and reported that multiple polling experts were critical of the concept. A NY Times article claims Rasmussen has a "record of relying on dubious sampling and weighting techniques." Rasmussen has also been criticized for only polling Likely Voters which, according to Politico, "potentially weeds out younger and minority voters".

A 2017 article by Chris Cillizza for CNN raised doubts about Rasmussen's accuracy, drawing attention specifically to potential sampling biases such as the exclusion of calls to cell phones (which, Cillizza argued, tended to exclude younger voters), and also more generally to a lack of methodological disclosure. Cillizza did, however, note in the same piece that Rasmussen was one of the more accurate polling organizations during the 2016 United States presidential election {sure, now & then they roll a 7}.

A Dec 2018 article by political writer and analyst Harry Enten called Rasmussen the least accurate pollster in the 2018 midterm elections after stating Rasmussen had projected the Republicans to come ahead nationally by one point, while at the time Democrats were actually winning the national House vote by 8.6 points—an error of nearly 10 points.

The Associated Press has also addressed Rasmussen's methodology. In 2018, AP journalists noted that Rasmussen's telephone methodology systematically omits adults, many of them young people, without landlines. The AP also noted that Rasmussen does not provide details regarding its online-panel methodology.

In an article for The Hill titled "Rasmussen Research has a pro-GOP bias," panelist discussed Rasmussen's practice of adjusting results by party identification. In addition to providing professional criticism from Ipsos, the article cited methodological concerns from Frank Newport of Gallup.
https://en.wikipedia.org/wiki/Rasmussen_Reports#:~:text=2020-,For%20the%202020%20United%20States%20presidential%20election%2C%20Rasmussen%20Reports'%20final,the%20election%20by%204.5%20percentage

I should add that Real Clear Politics is certainly aware of Rasmussen's outlier label, and that Rasmussen is included in every single RCP presidential approval poll, whereas most reputable polls are only included in about half, as they drop out of the polling average drop-back after about 2 weeks, since they only conduct approval polls about once a month.