Friday, January 9, 2009

Project Implicit is complete garbage

After reading a CNN article titled "You may be more racist than you think, study says," I followed a link in the article to Project Implicit on Harvard University's website, where you can supposedly take tests to discover your hidden biases regarding things like race, religion, age, weight, etc.

So far, I have taken three of Project Implicit's tests, and I am not impressed. First, I took the Black-White racial preference test, which said I have a strong automatic preference for White people. This apparently put me in Ku Klux Klan territory, as it was the most racist option possible.

Second, I took Project Implicit's Obama-McCain test, which actually tested two things: racial preference and politician preference. Regarding racial preference, it said I have no automatic preference for either Black people or White people, which was in complete disagreement with the racial preference test I had just taken a little while earlier. Regarding Obama vs. McCain, it said I have a slight preference for Obama. Although I did vote for Obama, I actually like both men but favored McCain most of the year. My switch from supporting McCain to supporting Obama was due to my dislike of Sarah Palin, not McCain, himself.

Finally, I took Project Implicit's "Presidents" test, which supposedly compares how one likes George W. Bush compared to other recent U.S. Presidents. For some background, I have despised George W. Bush as long as I have known about him. I contributed to McCain's 2000 campaign, in an attempt to prevent Bush from winning the Republican nomination. I voted for Gore in 2000 and Kerry in 2004. Even after the September 11 attacks, when Bush's approval rating soared to around 90%, I was in the other 10%. I think he is one of the worst presidents in U.S. history. So what did Project Implicit's test say? (You know where this is going, don't you?) It said I have a strong automatic preference for George W. Bush, which is complete garbage.

I don't know what research Project Implicit's tests are based on, but from my experience the tests are no more accurate than random chance. The New York Times has also noted that "there isn’t even that much consistency in the same person’s scores if the test is taken again." Yikes! This is science? Wikipedia lists some of the criticisms of their testing methodology here. In fact, if you search for the term "Implicit Association Test," which is the testing methodology Project Implicit uses, you keep coming back to the same very small group of researchers. This suggests that their methodology has not passed the peer review necessary to verify that it is reliable. I get the uneasy feeling that Project Implicit's tests are as bogus as polygraphs. You've gotta love social science research.


  1. We are genetically preprogrammed to be racist and xenophobic. To kill what we don't understand, or anyone who is different. To be otherwise would mean your ancestors would not have survived, and you could not exist.

    To state that you are not racist or biased to others is a lie. Everyone is.

    It is up to you, to use your intelligence to overrule these outdated instincts. We must all become aware of these feelings we naturally feel in order to overcome them.

  2. I think you are right that Project Implicit is complete garbage. The order in which they are asking the questions determines the supposed bias. As you learn to "play the game" you automatically begin to play faster, then it sets up the assumed "implicit association" and then determines that because you're playing faster you have the implicit bias. They could have accounted for this error and randomized the order in which the associations appeared, but they did not. Rather they ordered it to get the results they expected to get. This project is a complete waste of research money. They can't even begin to test whether our automatic reflexes are tied into our subconscious biases, because their testing method is so full of errors. Have these guys even taken a class in statistical analysis? RIDICULOUS!!

  3. I agree. The test "sets" one up to get the data that is wanted. Biased.