When the AP released a poll last week claiming that Romney had won over women voters there was a lot laughing.  Few took the poll seriously and new and old polls have debunked to the poll.

However, that didn’t happen when a second even more misleading AP poll was released the same day.  The second poll claimed Americans had become more racist and those headlines, of course, were spread across the globe, with claims that once again there is more proof that the U.S. is a terrible racist country.

The AP story also had a political angle.  Many news stories claimed a lack of leadership on race issues by the President’s caused the increased racism when most other reports suggest racism has increased within the Republican Party.  However, this alternative view was completely ignored in all the stories I reviewed.

Below, I put in quotes around comments from the researchers survey report and then below that in Italics I put what was reported in the press.

The scientific report (pdf) of a series of three surveys on racism was released a week ago.  The surveys were conducted in 2008, 2010 and 2012. The report was written by Josh Pasek, Fellow University of Michigan, Jon A. Krosnick, Assistant Professor, Stanford University and Trevor Tompson, NORC at the University of Chicago.

The researchers’ conclusions were

“The influence of racial attitudes on approval of Mr. Obama’s job performance decreased between 2010 and 2012.”

There were other problems in the study that prevented the a reasonable person from claiming that the percentage of overall racism has increased.  One problem was that the sample appeared to be older, more conservative and had a much higher percentage of Birthers than other national surveys. Only 49% said the President was born in this country which is the lowest percentage reported in any poll I could find in the past year. Recent polls put the number at 67%. The same poll put the number actually stating that the President wasn’t born in this country at 15%. However, the researchers’ survey put the number at 39%! In other words, the sample was severely biased toward Birthers and may have had a greater percentage of racists that other surveys. This bias wasn’t reported in any of the news articles.

There was more evidence of a conservative sample. A slightly higher percentage of the country is liberal. For the survey sample, respondents identified themselves as more conservative (another fact not reported in the press)

Politics 2010 2012
Very liberal 8% 5%
Somewhat liberal 17% 15%
Moderate 40% 41%
Somewhat conservative 23% 24%
Very conservative 11% 10%
Not Answered 2% 5%

 

“Converting both anti‐Black and pro‐Black attitudes to neutral led to a projected increase in Mr. Obama’s 2012 vote share of 2 percentage points and a projected decrease in Mr. Romney’s 2012 vote share of 3 percentage points.” Overall, racism on an explicit measure of racism increased over the last two years (47.6% in 2008; 47.3% in 2010; 50.9% in 2012).

This report is good news for the President.  Prior research on the 2008 race argued that racism cost President Obama more votes (about 7% of all votes) 3,4,5 as opposed to the 2% estimate of the current study.

The Press Reported the Influence of the 2012 Election will be more influenced by racial attitudes.  No news reports mentioned the sample was more older, more Birther orconservative.  However, the AP cut off commenting when commenters began pointing out the problems with the survey.  No reports claimed that the influence of racism decreased which was the researchers opinion.  Rather the reports stated the opposite of the researchers conclusions claiming racism was rampant among All Americans since Obama had become President four years ago.

The researchers found the average in their measure of explicit prejudice didn’t increase between 2008 and 2010 but did increase between 2010 and 2012 (47.6% in 2008; 47.3% in 2010; 50.9% in 2012). However, in terms of actual percentages, there were no statistically significant differences over time .

The Press stories emphasized that racism increased over the last four years (not two years as the data suggest) when reporting the statistically insignificant differences that may have been a result of sampling error.  Many foreign newspapers ran with the same headlines as the U.S. newspapers reporting that Americans had become more prejudiced as a result of the Obama presidency.

The researchers also said

“People who identified themselves as Republicans in 2012 expressed anti‐Black attitudes more often than did Republicans identified in 2008.”

There were no differences found for Independents or Democrats.

No one in the press reported this finding which is statistically significant.

The report also said

“In 2008, the proportion of people expressing anti‐Black attitudes was 31% among Democrats, 49% among independents, and 71% among Republicans.”

In 2012, the proportion of people expressing anti‐Black attitudes was 32% among Democrats, 48% among independents, and 79% among Republicans.”

Most reports didn’t report these findings.  None reported that the trend was statistically significant for Republicans.  Instead, a few reported the difference was slight.  

Think about that for a moment.  A 3% difference in the overall population between 2008 & 2012 is the main theme of all articles I examined (over 100) even though it could just be a result of survey sampling error.  However, a statistically significant different of 8% is glossed over or not mentioned in all 300+ reports.  The obvious conclusion is that Republicans became more prejudiced and Independents and Democrats didn’t and the researchers noted that these finding are consistent with other research.  However, no one in the press reported that obvious conclusion. 

For an implicit measure of racism, the report said

“In 2012, the proportion of people expressing anti‐Black attitudes was 55% among Democrats, 49% among independents, and 64% among Republicans”

These finding were reported in the press to illustrate that Democrats were almost as prejudiced as Republicans.  The news articles that did this failed to report the outcomes for the explicit measure.

However, News Busters reported that the implicit measure obviously didn’t work.  The Implicit measure is supposed to tap racism you’re willing to admit to plus racism you’re not willing to admit so the percentages should be higher for the Implicit measure of racism and they are for Democrats.  However, they are lower for Republicans so clearly there is something wrong with a measure that can’t tap all of the overt racism, let alone covert racism.   

The researchers said

“Implicit racial attitudes were measured using the Affect Misattribution Procedure (AMP). Respondents saw a series of Chinese ideographs, one at a time, and assigned each one to one of two categories, more pleasant and less pleasant, placing approximately half of the ideographs in each. After respondents completed a sequence of practice trials during which only ideographs were shown, more ideographs were shown, but each one was preceded by a quick flash of a photograph of the face of either an African-American male or a White male (24 of each were used). When a face precedes an ideograph, people’s affective reactions to the face influence their later assessments of the ideograph (Payne, Cheng, Govorun, & Stewart, 2005). People who have favorable feelings toward the face are more likely to label the ideograph as pleasant, and people who have unfavorable feelings toward the face are more likely to label the ideograph as unpleasant. Two scores were calculated for each respondent: the proportion of trials rated as pleasant when an African-American face was shown, and the proportion rated as pleasant when a White face was shown. We subtracted the score for trials with African-American faces from the score for trials with White faces and then recoded the result to range from 0 (meaning all ideographs preceded by African-American faces were labeled as more pleasant and all ideographs preceded by White faces were labeled as less pleasant) to 1 (meaning all ideographs preceded by African-American faces were labeled as more pleasant and all ideographs preceded by White faces were labeled as less pleasant). “

Recent research suggests there may be problems with this measure.  In fact, it may be inversely correlated with prejudice among people with liberal attitudes and there appears to be additional technical problems as well.

The researchers also reported that if the election were held in August when the survey was administered, President Obama would win the election.

“Given current actual levels of racism, the model predicts that the President would receive 47.1% of the vote and that Mr. Romney would receive 41.8% of the vote.”

However, the researchers noted

“Many events have happened since then and will continue to occur as we approach election day, and these events may alter the impact of anti‐Black attitudes.”

The Press disregarded the researchers warnings and interpreted the findings of the study as suggesting that racism in both parties could cost Americans the election.  In other words, the press’ conclusion was oblique to the researchers’s conclusions.

Jennifer Agiesta, Sonya Ross & Dennis Junius wrote stories for APNews.

The story has gone unchallenged by the mainstream media for a week now and AP has refused to retract the story.  The definition of the mainstream media here is broad and includes liberal and progressive news, start-up news websites, websites focusing on African-American and Hispanic audiences and liberal websites.  Below is a sample newspapers that have refused to retract or refused to respond to requests for retraction of the story

All of these sources have had several days to retract their stories (except for SF which was just contacted this afternoon).  Although some admitted to me the story was false, not one news organization retracted the story or ran a counter story.

The conservative press either ignored the story or came up with a conspiracy theory that President Obama, the AP and the survey company were attempting to blame a forgone election loss on racism ( Limbaugh ) or argued that the surveyors were psychologically forcing respondents to report false stereotypes ( Wall Street Journal ).

In my experience with this story and other media stories, news organizations have no concern or means to self-correct and no oversight.  A somewhat more detailed report of the problems with the news reports can be found here.