Friday, June 15, 2012

When politics drives research you get neither good research nor good politics

Harvard ‘racism’ researcher neglected to disclose White House ties

The Harvard researcher behind a recent study correlating 2008 election results with racially charged Google searches neglected to disclose ties to a former senior member of the Obama administration, The Daily Caller has learned.

Seth Stephens-Davidowitz, an economics Ph.D. candidate, lists on his C.V. — publicly available through his website — that he was the research assistant of Peter Orzsag at the Brookings Institution from August 2005 to August 2006. Set to graduate in 2013, Stephens-Davidowitz entered the Ph.D. program at Harvard in 2007.

Orszag, currently the vice chairman of Global Banking at Citigroup, headed the Office of Management and Budget as an Obama appointee between January 2009 and July 2010. He was also the director of the Congressional Budget Office from January 2007 to November 2008.

Whether this relationship between Stephens-Davidowitz and Orszag affected the outcome of the study, either directly or indirectly, is uncertain. Neither Stephens-Davidowitz or Orszag returned The Daily Caller’s request for comment.

“To my knowledge, Mr. Orszag has no idea of the contents of the study, nor do any of his former colleagues,” professor David Cutler, an adviser to Stephens-Davidowitz at Harvard, told The Daily Caller. “Seth wrote this paper at Harvard.”

Stephens-Davidowitz also thanked “three anonymous referees” along with various named sources in the first footnote of his study. Cutler, however, said that he believes Stephens-Davidowitz “appropriately thanks people who gave him advice on his paper.”

“Some people prefer not to be thanked, and he needs to respect their right to privacy,” Cutler added.

Using Google Insights for Search, Stephens-Davidowitz charted the percentage of searches between 2004-2007 that included the terms “nigger” or “niggers” to gauge racist sentiment. Google Insights for Search is a program made publicly available by Google to allow users to gain comparative insights into search terms in various markets since 2004.

“I choose the most salient word,” he wrote in an explanatory piece in the New York Times. “I do not include data after 2007 to avoid capturing reverse causation, with dislike for Obama causing individuals to use racially charged language on Google.”

He then compared his results against the 2008 election results in various areas, explaining that the “conditions under which people use Google — online, most likely alone, not participating in an official survey — are ideal for capturing what they are really thinking and feeling.”

He found that areas where Obama “underperformed” — West Virginia, eastern Ohio, western Pennsylvania, upstate New York and southern Mississippi — were the regions with the highest rates of racially charged searches.

“If my findings are correct, race could very well prove decisive against Mr. Obama in 2012,” Stephens-Davidowitz said.

He also compared exit poll data between Massachusetts Democratic Sen. John Kerry from 2004 and Obama in 2008 to examine relative preference for each candidate by race.

“If states with high racially charged search were more likely to support Obama, independent of white’s racial attitudes, the effect would likely show up for both black and white voters in these states,” Stephens-Davidowitz wrote in the paper.

“Instead, there is no relationship between racially charged search and black support for Obama relative to Kerry; the relationship is entirely driven by white voters,” he continued.

In a footnote, however, he admitted to painting with a broad brush.

“Throughout this paper I refer to non-blacks, including Hispanics and Asians, rather imprecisely, as ‘whites,’” he said.




No comments: