A method and system applies a rating adjustment to ratings by avatars in a virtual universe. The rating adjustment offsets mutual rating collusion between avatars. The ratings may be adjusted based on a weighted value of indicia of association, and may also be adjusted based on rating frequencies between the same avatars. In a rating system having a fee structure, the ratings may be adjusted by scaling the rating/fee payoff to favor higher fees. Where indicia of association are weighted, the weightings are mapped to discrete levels of association between avatars, from no association to close association, with greater reduction of ratings for closer association.