The second approach was an age-period design [31]: We compared the prevalence of substance use among ABCD Study participants who were 11 or 12 years old in the years 2018, 2019, or (May/June) 2020 (total n = 7,585 11-year-olds, 3,549 12-year-olds). The rates of substance use can be conceptualized as varying along three dimensions: age, period (i.e., calendar year), and cohort (i.e., year of birth) [31]. If we hold age constant (e.g., compare prevalences only within 11-year-olds) and assume no cohort effects (see Supplement for justification), then differences in prevalence can be attributed to the calendar year in which the assessment was completed. This design is possible in the ACBD Study because recruitment was rolling (2016–2018), so although all youth were aged 9–10 years at study entry, participants varied in age in any given calendar year. Inspection of the distribution of youth age across calendar years revealed that we had substantial numbers of qualifying prepandemic assessments of 11-year-olds in the years 2018/2019 and of 12-year-olds in the year 2019. Thus, the age-period design consisted of comparing the prevalence of substance
