Mail questionnaire response rates were investigated in a three-factor design including ethnic group (Spanish surname vs. non-Spanish surname), income level (below vs. above median income) and questionnaire language (English only vs. English with interlinear Spanish translations). Log-linear multiway frequency analysis indicated lower response rates were received from addressees who were Spanish-surnamed, below median income and who had received bilingual questionnaires. Non-significant two-way interactions suggested that Spanish-surname addressees had relatively lower response rates for both income levels and that the translation manipulation lowered response rates equally for Spanish surname and non-Spanish surname addressees. The argument that the Spanish translation questionnaire may have "appeared" longer and hence reduced response rates is refuted on the basis of available literature reviews which indicate that length of questionnaire is not influential in reducing response rates. The problems of increasing response rates from Hispanic consumers were discussed. (Author)