Instability Overlooked: Evidence for the Importance of Household Roster Data Collection and Matching over Time

Abstract

Many studies on instability examine changes in relationships and household composition by using measures such as “are you married?” or “how many children in this household?” and then comparing the answers across time. Instability is determined from changes noted in respondents‟ answers and conclusions are made based on the impact of instability, or the lack thereof. However, these measures are one-dimensional; they do not capture the types of changes that may have occurred between waves. For example, respondents may be married at both waves, but to different people. This misconception calls for a more nuanced examination of households and relationships of members at each point in time. Data from the Making Connections Survey are used to compare different methods of measuring instability in relationships, number of adults, and number of children across waves. The data include a household roster that collects demographic information about each member of the household, as well as their relationship to the respondent. Moreover, household members on the roster are matched across waves and each person is given a unique identifier. This makes it is possible to tell changes in household composition and relationships on an individual basis, instead of relying on singular measures. Findings show that when longitudinal studies closely examine household rosters and link household members over time to measure instability, instead of using only single measures - a significantly greater proportion of change is captured. This study is important because it illustrates the importance of more detailed data collection on household members, and considering the fluid nature of especially poor families, the extent to which instability is overlooked by the usual measures. Finally this study provides evidence suggesting that conventional methods of capturing instability related to people in a household should be revisited.