I took this demo test on this site and am gratified to say that my result was:
Your data suggest little to no automatic preference between Other People and Arab Muslims.
These tests are weird. For one thing, the instructions are complicated. For another thing, it seems that they spend the first half conditioning you to associate Arab names with bad things, disaster, failure, etc. Then you have to disassociate them in your mind afterwards. I was a little worried that my results would show me as biased, after that. But *whew*, they didn’t.
Now that I’ve read more, apparently people get different versions, so somebody else’s test might have had them associating Arab names with good things first.