Two issues here Takeda.
First - the "his methodology is excellent" comment. C'mon now - your smarter than this!
1) He chooses as his search criteria the "N" word - which is "racially charged" and derogatory.
2) He then disregards the leading source of the word - rap lyrics. This alone should toss the study - because if the word is derogatory - who seems to be using it the most? He has to take it out - or else his "study" would show that the most racially biased group against African Americans is - African Americans!
Since that can't be right, he just ignores it.
3) Next he takes the modified, aggregate data and determines that the rest of the searches lead to primarly derogatory material about African Americans.
4) He then proceeds to 2008 - and finds "Obama" was "one of" the most searched items that were "racially tinged". Notice the difference in wording here - racially charged is a negative - racially tinged is not necessarily the case. A person searching "Obama Indian" would have created a "racially tinged" search. Yet there is nothing inherently racially baised in such a search. It would be the same as searching "Obama Hawiian". His choice to make anything speaking to race equate to a form of racism is a severe technical flaw.
5)At no time does he have data demonstrating the race of the searcher. His conclusion of racism based purely upon the search term makes a clear assumption regarding the race of the searcher. For such racism to be inferred, he must also assume that the searchers are all non-african in descent.
To come to the conclusion that he did - the study implied that racism cost Obama 3.1 to 5.0 percentage points in the last election - is absolutely without basis. Why is issue 2.
1) There is no correlating data that demonstrates how many "racist" searchers cast a vote.
2) There is no supporting data that shows who such "racist" searchers voted for.
As he states - the mere "racially tinged" search "implies" racial hostility - a flawed implication to start. He then takes that and concludes that such racial hostility "implies" a significant cost to Obama in the previous general election.
A conclusion build on an implication of an implication that ignores some data and assumes other data not available - with "standards" like that, why isn't this guy researching global warming???? He could be getting major government grants with practices like that!