Cohort-Sequential (Accelerated Longitudinal) Design

Description

Follow several entry cohorts at once to cover a long span of age or tenure quickly, adding fresh participants at each wave to estimate retest effects.

Sequential designs separate three confounded influences: age (or time since an event), cohort and period. A cohort-sequential design follows several cohorts that start at staggered points of the span of interest, measuring all of them at common calendar waves, so a long span is covered in far fewer years than a single-cohort study. Overlapping segments are linked to estimate a single trajectory. Adding a fresh, previously unmeasured sample at each wave, a retest-control or refreshment cohort, lets researchers estimate how much repeated measurement itself changes responses. Design cautions include using change-sensitive measures, remembering that measurement invariance does not guarantee identical meaning, and choosing the time metric and interval deliberately, since annual waves miss faster processes.

Attributes

Loss type
Unmigrated
Effort
Months

Steps

  1. Define the span of interest in age, tenure or time since adoption.
  2. Recruit several cohorts that start at staggered points of that span (for example new users, 6-month users and 12-month users).
  3. Measure all cohorts at common calendar waves with change-sensitive measures.
  4. At each wave add a fresh, previously unmeasured sample to estimate retest effects.
  5. Model time-in-span, cohort and period effects and compare fresh with repeat respondents.
  6. Link overlapping cohort segments into a single estimated trajectory, checking invariance across cohorts.

Sources

  • Handbook of Child Psychology Vol 1 (Lerner et al., 2015)
  • A General Model for the Study of Developmental Problems (Schaie, 1965)

Also known as

  • Cohort-sequential design with retest-control cohort
  • Accelerated longitudinal design
  • Sequential designs (Schaie)
  • Refreshment sample

Categories

  • Research Operations
  • Research Design & Rigor