Because item order and salience may affect the findings of social science research, a study was conducted to determine the effect of topic salience on subject response to different item orders. In the study, 10 high salience self-esteem items were presented with 10 low salience items concerning product labels in three versions. In the first version, all the self-esteem items were followed by all the label items; in the second version, all the label items were followed by all the self-esteem items; and in the third version, the 20 items were scrambled. Subjects, 128 college students enrolled in a course in human communication, rated the items on a five point likert-type scale from "very unimportant" to "very important." Only those persons who perceived self-esteem to be more important than product labels were used in the analysis of the data. Results indicated that the more deeply held an attitude, the more salient it may be regarded to be and the less likely it is to be affected by any surrounding questions; issues of only passing concern, however, may be strikingly affected by the surrounding context of questions. These findings indicate that extreme care should be taken in interpreting responses of people with little interest in the issue at hand, that data should be factor analyzed, and that respondents should probably be asked how important an issue is to them whenever a questionnaire is administered. (JL)