Measurement invariance is important when attempting to compare groups on underlying latent characteristics like nicotine dependence. Measurement non-invariance can lead to incorrect conclusions about the differences across groups, obscuring differences that exist or creating differences due to measurement non-invariance rather than actual differences, as well as add heterogeneity to analyses that combine cohorts from different eras (e.g., meta-analytic genome-wide association studies). For the first time, we assessed FTND measurement invariance by birth cohort using cross-sectional studies that collectively comprise individuals born from 1926 through 1998. The pattern of responses to current FTND items has changed over time, including higher CPD and lower tendency to smoke when sick in the oldest respondents (birth cohort <1945). Thus, verifying measurement invariance across different birth cohorts is important for instilling confidence in study results that combine or compare FTND across different age groups.
