Posted on behalf of Rachel Courtland:

We’d like to think we’re making informed, balanced decisions when it comes to elections. But what if we’re not? In December, University of Washington researchers released preliminary findings showing a distinct disconnect between “unconscious” and explicitly declared preferences for presidential candidates. While 42 percent of test-takers said they would vote for Senator Barack Obama in the Democratic primary, the proportion dropped to 25 percent in a variation of Implicit Bias Test — a test that, since its introduction in 1998, has measured what researchers say are nearly universally-held biases, particularly in matters of race and gender. Senator Hillary Clinton fared better in the test for unconscious preferences, taking 48 percent of voters when only 34 percent explicitly said they favored her.

The new test, which you can take here, works by measuring tiny variations in how long it takes a subject to sort words into categories like “good” and “bad” while being shown images of presidential candidates.

The test is not universally loved among social scientists, as the results may not correlate with any real-world behavior. And Harvard psychologist Mahzarin Banaji, who helped develop the test, worries over her own results. In a recent commentary in the Atlanta Journal-Constitution, she notes that the results tell her “that I may not be fully aware of who I am or wish to be. What I take away from such a fracture in my own mind is a skepticism that I am color-blind or that I can look past gender to the truly competent candidate.”

Be forewarned: the test is long, so set aside some time to take it. I didn’t get through it all the first time.