Okay, I'll bite, and I won't even use the word regatta in my example since y'all are turning your nose up at it.
This is from Ji, Zhang, and Nisbett, 2004.
Let's say on a children's intelligence test they were given a list of words and asked to pick the two that "go together", a pretty straight-forward task:
A. Seagull
B. Squirrel
C. Tree
A. Monkey
B. Panda
C. Banana
A. Cow
B. Milk
C. Pig
A. Foot
B. Shoe
C. Hand
European American students are more likely to group based on taxonomic categorization: seagull and squirrel, monkey and panda, cow and pig, foot and hand.
East Asian students are more likely to group based on thematic categorization: squirrel and tree, monkey and banana, cow and milk, foot and shoe.
Of course, when people are talking about cultural biases in testing, they're usually less concerned with testing children. You'll see some of it in admissions testing, but a lot of the problems revolve around employee and personnel selection. Spearman's
g, for instance, is a general intelligence measure that typically correlates highly with work performance, but using it will cause adverse impact for pretty much anyone who isn't a white male.