Studies on the stereotype threat have typically used stereotypes that that are common in contemporary culture, i.e. the canonical example being the study that showed African American SAT scores lowered when faced with the stereotype that African Americans are less intelligent than other groups.
Have there been any studies where researchers invented a stereotype (i.e. something random, like "Hispanics perform more poorly at jump rope than other groups") and tested to see if it had any effect on well controlled test subjects?
Have there been any social studies where scientists have tried inventing a new stereotype and implanting it into the mainstream population? (I doubt this because of ethics issues, but it would certainly be interesting!)