Part-whole combinations of questions are believed to be particularly vulnerable to question-order effects. A split-ballot experiment was used in a telephone survey to vary the order of a general well-being question and a set of questions on well-being in eight specific life domains. A number of order effects on multiple regression parameters were found for married respondents; no order effects were found for unmarried persons, however. These results show that conclusions about the relative importance of specific life domains for general well-being and conclusions about the effects of unmeasured variables can be substantially affected by the order of the general and specific items. Several explanations for the order effects are discussed.