Survey research in New Mexico Hispanics: some methodological issues

A prevalence survey of respiratory diseases was conducted in Albuquerque, New Mexico, with the objective of explaining differing patterns of respiratory disease epidemiology in Hispanic and non-Hispanic whites (Anglos). The study population was selected at random from the 1978 R.L. Polk & Co. Directory. This paper focuses on methodological issues raised during the conduct of the study: response rates, potential language barriers and bias, and identification of Hispanics by surnames. Mail, telephone, and personal interview approaches were used to obtain adequate response rates, which ranged from 60% in Hispanic males to 78% in Anglo females; 22% of Hispanic males refused interview. Fewer Hispanics returned mailed questionnaires than responded to telephone interviewing. Spanish language was increasingly preferred as the respondent's age increased. Two methods of ethnic identification by surname (1980 Census List of Spanish Surnames and a computer program, GUESS (Generally Useful Ethnic Search System) were compared to the self-reported ethnicity of respondents. The GUESS Program was more sensitive than the census list, but the census list was more specific. The combination of both methods produced a 90% sensitivity and 97% specificity in males. Intermarriage reduced the accuracy in females.