The Intrahousehold Communications Study: Estimating the Accuracy of Proxy Responses at the Dyadic Level

What do people living in the same household know about each other? Do they, for example, know the employment status, work and vacation hours, and wages of other household members? Similarly, do they know what other household members spend money on and how much they spend? Although there is a growing body of literature on what household members serving as proxy reporters actually know about other household members, and on the factors that might account for any observed differences between self and proxy reports, the results are rather inconclusive (e.g., Moore, 1988). In brief, observed differences between self and proxy reports seem to be, at least in part, related to factors such as the relationship between the proxy and the self, the information being asked about, the importance of the information to the proxy, the characteristics of the proxy, whether the proxy participated in the event, and the method of survey administration (see for example Blair, Menon, & Bickart, 1990; Cash & Moss, 1974; Groves, 1989; Mathiowetz & Groves, 1985; Moore, 1988; Rodgers & Herzog; 1987; Rodgers, Herzog, & Andrews, 1988). Nevertheless, many surveys continue to accept one individual's report about all other household members because it is neither reasonable to expect nor practical to seek individual reports from each person in every household (e.g., Blair, Menon, & B ickart, 1990; Dippo, 1989). We had two specific goals in this research. Our first goal was to examine the accuracy of proxy reports involving household members' expenditures. We employed survey questions similar to those for which the Bureau of Labor Statistics (BLS) frequently accepts proxy reports. Our second goal was to identify factors that would be systematically related to the accuracy of these proxy reports and that could subsequently lead to more efficient and effective selection of adequate proxy respondents. Proxy Accuracy Although self-reporters are fallible human beings who may forget some of their expenditures or may include expenditures made outside of the reference period, they are probably the best criteria that can be reasonably obtained for studying everyday expenditures, and they also represent the best alternative to proxy reports for survey researchers. In the present investigation we compared proxy reports to self reports and utilized their degree of agreement (or disagreement) as our criteria for examining potential predictors of proxy accuracy. Predictors of Proxy Accuracy A variety of factors may be related to how much one person knows about another household member. There are likely to be some individual characteristics that distinguish some people, who by virtue of their role or status in the household, are likely to be well informed about the expenditures of other household members. Some people may also have a great deal of knowledge about particular household members or may learn information about other household members as a result of the type or amount of direct or indirect interaction they have with other household members. Furthermore, some people may know quite a bit about a particular household member because of their relationship or their interest and involvement in the activities of that particular household member. At yet another level of analysis, there may also be household or family characteristics that distinguish households where all members are well-informed about each other from households where each person can only be relied upon to accurately report about themselves. In this paper, we focus on the dyad, looking at the direct and indirect interaction of the self and proxy or the source or mode by which the proxy learns about other household members (through direct participation, communication, or observation). We also look at the relationship between the proxy and the other household members they report on. Mode or source of learning. At the most basic level, what one person knows about another must be learned through some kind of direct or indirect interaction. This interaction could be very informative and direct, such as participating in the same activity together, or discussing a purchase or work schedule. For example, Mary may tell John how busy her work schedule will be next week, or she may go with him to buy groceries. This interaction could also be indirect, such as John noticing that Mary is wearing a new blazer or that she has come home late from work a couple of nights in a row. Of course, one person may learn about another by interacting with a third person who had a more direct interaction with the second person. Julie may, for example, ask her father about