changes. In addition to examining changes within different cohorts over time, the interrelationships among the lineages can be simultaneously examined. Lineages are represented by their horizontal interrelationship within time of measurement. The age overlap within generations is shown; as Hagestad ( 1981) indicated, family members do not file in orderly ways into generation lineage positions by cohort and age. Because of the proposed probability sampling approach to lineages ( Jackson & Hatchett, 1986), individuals within lineages have individual probabilities attached to their selection. Thus, they can be followed as longitudinal cohort sequences over time ( Baltes et al., 1979). Similarly, within time of testing, for some purposes samples can be treated as cross-sectional designs, the most typical way in which lineage data have been treated in the past ( Hagestad, 1981; Jackson & Hatchett, 1986). Finally, the true panel nature of the cohorts across age can be utilized and the changes in lineage interrelationships assessed across time. The proposed multigeneration design incorporates in one package many of the conceptual advantages claimed for cross- sequential research approaches and designs ( Baltes et al., 1979; Nesselroade & Labouvie, 1985; Schaie & Hertzog, 1982, 1985).
We know of only two large, multigenerational studies that have included a longitudinal component. The first was the pioneering study of Hill ( 1970) that followed lineages over a 7-year period. The other is the study by Bengtson ( 1985), which followed multigeneration lineages over a 15-year period ( Mangen, Bengtson, & Landry, 1988). In the Hill ( 1970) study the generation linkages were not fully exploited in the longitudinal sequences. In all fairness to Hill, the types of analytical procedures most appropriate to this type of study (e.g., event history analysis, structural equation modeling, two-stage least squares, path analysis, etc.) were not widely applied in the social sciences when his study was conducted. Longitudinal analyses on the Bengtson ( 1986) study are currently in progress. Although not specifically designed with this approach in mind, the Panel Study of Income Dynamics ( Duncan & Morgan, 1985) incorporates some of these features as well.
As shown in Fig. 4.1, because of the nature of the lineages, cohort position of the lineage members, entrance of new members over time, and the opportunity of observing reciprocal influence with multiple occasion measures, the Multi-generation Family Design offers a unique opportunity for studying individual and family development.
No particular research design is a panacea for the investigation of long-term changes and stability in human development ( Glen & Frisbee, 1977). The