WEEK 8- DEBATE READING SUMMARY
Reading 1- Notes on ‘Drinking is Fun’ and ‘There’s Nothing You Can Do About It’: The Problem With the 21-Year-Old Minimum Drinking Age
Overview
Editorial by Reginald Fennell (2007) in Journal of American College Health examining the 21-year-old minimum drinking age in the United States and arguing it may be time to re-evaluate this policy.
Central claim: prohibition by age (minimum drinking age) has not eliminated underage drinking on college campuses; a more holistic approach to alcohol culture and safer drinking practices may be needed.
The piece blends personal narrative, policy history, data from national agencies, and suggestions for campus-level and national policy changes.
Background and Context
Historical backdrop:
Prohibition in the early 20th century failed to eliminate alcohol consumption.
National Minimum Drinking Age Act of 1984 effectively forced states to set the minimum drinking age (MDMA) at 21 to receive federal highway funds.
The law is backed by evidence that it reduced fatalities from drinking and driving, but it contains exceptions (some states allow drinking under 21 with parental consent; parental presence varies by state).
International comparison:
Many countries have a minimum drinking age of 18; some allow 16-year-olds to drink.
The U.S. policy of 21 remains unique and controversial in light of international norms.
Underage drinking on campuses persists despite policy: various authors in the issue document ongoing issues, including risky behaviors and harm.
National statistics referenced:
NIAAA estimates that excessive drinking leads to more than student deaths, nearly injuries, and nearly assaults annually.
In 2003, of youth ages reported consuming alcohol in the past 30 days.
Policy mechanisms:
The MDMA’s impact on fatality reductions via reduced drinking and driving is acknowledged, but underage use remains common within the campus milieu.
Evidence and Data
Ignition interlocks and drunk driving:
National Highway Traffic Safety Administration (NHTSA) advocated stronger use of ignition interlocks for repeat drunk driving offenders; as of the cited date, about arrests occur yearly for impaired driving, with a third being repeat offenders; ignition interlocks are widely available in states (except a few).
College health and behavior data:
ACHA–NCHA data indicated that of students (out of a large sample) reported never drinking; excluding non-drinkers and non-drivers, reported driving after drinking in the past 30 days.
Healthy Campus 2010: goal to reduce alcohol/drug-related deaths and injuries from motor vehicle crashes.
Research themes and recommendations:
NIAAA task force on college drinking recommends keeping students busy (e.g., morning classes, more Friday classes) to reduce drinking avenues, though this may be imperfect given online course materials.
Pedersen and LaBrie highlight prepartying (drinking before going to a party) with high prevalence among underage students; their work notes a significant portion engage in drinking prior to events.
Howard and colleagues summarize qualitative findings: students resist being told to abstain; they want knowledge on how to drink responsibly (pacing, limits, taking care of others).
Consequences and policy tensions:
Despite the policy, significant alcohol-related morbidity and mortality persist among college students.
There is a desire to balance enforcement of laws against illegal drinking with harm-reduction strategies that teach responsible use.
Personal Background and Narrative Context
The author shares personal experiences to frame the discussion:
A recreational non-drinker who travels and works in France; background includes witnessing alcohol-related issues in a close friend’s family; intention to provide context rather than moral judgment.
A triathlon-related anecdote: a college freshman with alcohol on his breath during a 5:30 AM drive to a race, illustrating real-world campus drinking dynamics.
The article’s ethos:
Acknowledges the author’s own and others’ exposure to alcohol and drug issues, and uses these experiences to motivate a policy discussion.
Proposals, Arguments, and Debates
Core argument: reexamine the 21-year-old minimum drinking age and consider lowering it (to 18) or abolishing the 21-year-old age barrier, potentially replacing prohibition with more nuanced control measures.
Potential policy options discussed:
Lower the MDMA to 18 (or even lower in some European-style models) to align the legal status with adult status and reduce the punitive focus of prohibition.
Maintain enforcement against drunk driving but implement broader harm-reduction strategies (e.g., ignition locks for all drivers) to reduce alcohol-related crashes
Strengthen ignition interlocks for offenders to prevent DUI incidents; the logic is to prevent dangerous behavior before it harms others or oneself.
Educational and behavioral approaches on campuses:
Promote responsible drinking knowledge (how to pace, how to know one’s limit, how to care for peers).
Increase meaningful, engaging campus activities to reduce time available for excessive drinking (e.g., more scheduled activities, community service requirements).
Reevaluate the efficacy of expanding class schedules (e.g., early morning/Friday classes) as a universal fix, given increased online access to course materials and the ability to learn asynchronously.
Programs targeting athletes and other high-risk groups:
Brenner et al. suggest alternative activities for athletes during nonseason periods to reduce high-risk drinking.
Pedersen and LaBrie suggest addressing prepartying and high-risk drinking through education and targeted interventions.
Behavioral and cultural shifts:
Emphasize cultivating a culture of responsibility rather than solely enforcing abstinence.
Implement service requirements (monthly community service hours) to foster citizenship and engagement, potentially increasing students’ sense of responsibility.
Vincent Clinton’s Giving (as cited) promotes giving as educational and potentially transformative, suggesting that engagement can have lasting positive effects.
Addressing the “drinking is fun” paradigm:
Howard and colleagues note that students perceive drinking as an ingrained campus culture; they desire practical knowledge on safe consumption, not just abstinence messaging.
The editorial uses the refrain “drinking is fun” and “there’s nothing you can do about it” as concerns to challenge, arguing for proactive, evidence-based strategies.
Practical Implications and Implementation Challenges
Enforcement vs. education:
While laws (e.g., MDMA) and DUI penalties matter, they do not fully deter underage drinking or its harms on campus; a combination approach is needed.
Campus responsibility:
Colleges could shift from prohibition to fostering responsible behavior, while recognizing legal limits and protecting student safety.
Unintended consequences:
Changes such as earlier class times or more online materials could inadvertently shift alcohol consumption patterns, possibly concentrating intoxication at different times; need to monitor outcomes.
Feasibility of radical measures:
Abolishing the 21-year-old minimum and introducing universal ignition locks would require substantial societal, political, and logistical changes and face significant resistance.
Cultural and Ethical Dimensions
Tension between autonomy and public health:
If the policy changes, should adults be allowed to drink at 18 with more permissive controls, even if it might increase harm in some scenarios?
The piece questions whether prosecuting someone who has had a drink but caused no harm should be avoided, reflecting broader debates about the boundaries of legal responsibility and moral hazard.
Normalization of alcohol and the “fun” narrative:
The repeated assertion that “drinking is fun” highlights how social norms shape behavior and the difficulty of changing campus culture through policy alone.
Illustrative Case Illustrations and Anecdotes
Triathlon student example:
The freshman with alcohol on his breath caught in a 90-minute drive to a state park; illustrates competitive athletes who nonetheless engage in underage drinking.
Student reflections:
A classroom exercise where students submitted papers on “Drinking Is Fun,” with few acknowledging negative consequences, underscoring the perception gap between beliefs and risks.
Recurrent quotes cited by the editor:
“Drinking is fun”
“There’s nothing you can do about it”
The BRAD birthday cards (referenced in Hembroff et al.):
Cards mailed to students before their 21st birthdays to reinforce awareness of risk; used to illustrate the ongoing monitoring of underage drinking.
Relationship to Related Policies and Programs
NIAAA College Materials and Task Force:
“College Drinking—Changing the Culture” emphasizes structural changes and campus-wide programs rather than solely punitive measures.
ACHA and NCHA data:
The data illustrate current prevalence and risk, informing campus health initiatives and policy development.
National and state policy context:
Ignition interlocks and DUI enforcement are framed as part of a broader public health strategy to reduce alcohol-related harm.
Related Materials and Brochures (Page 5)
ACHA publications and pricing:
2006-2007 College Health Salary and Staffing Survey Report: data on salary ranges and staffing for health services, useful for institutional planning.
The report includes salary ranges (minimum, 25th percentile, median, 75th percentile, maximum) and staffing levels by institution type and size.
Prices: Member Institution $75 (non-participant) to $150 scale; individual member $90–$135; nonmember $105–$165 (illustrative student pricing from the page).
Brochure series for health topics:
Meningitis on Campus: Know Your Risk; Immunizations (HPV, etc.); Stress in College: What Everyone Should Know; Alcohol Use and You; Decisions on Tap.
Other topics include Eating Disorders: What Everyone Should Know and Nutritional Serving Size information.
Contact information for ordering:
Visit www.acha.org or email pubs@acha.org for catalogs or samples.
Key References (as cited in the editorial)
1. National Minimum Drinking Age Act of 1984, 23 U.S.C. § 158. (Policy basis and citation of the law.)
2. National Institute on Alcohol Abuse and Alcoholism (NIAAA). Surveillance Report #74: Trends in Underage Drinking in the United States, 1991–2003. (Underage drinking prevalence.)
3. National Institute on Alcohol Abuse and Alcoholism. Alcohol Policy Information System (APIS) Web site. (Policy history.)
4. National Highway Traffic Safety Administration. Ignition interlocks for repeat drunk driving offenders—press release, Aug 22, 2007. (Policy proposal for ignition locks.)
5. American College Health Association. The ACHA–NCHA Spring 2006 reference group data report (abridged). J Am Coll Health. 2007;55:195-206. (Student drinking behaviors.)
6. American College Health Association. Healthy Campus 2010: Making It Happen. (Strategic health objectives for campuses.)
7. National Institute of Alcohol Abuse and Alcoholism. College Drinking—Changing the Culture [NIAAA College Materials]. (Changing campus culture approach.)
8. Clinton WJ. Giving: How Each of Us Can Change the World. (Advocacy for giving and engagement as educational.)
Synthesis and Takeaways
The MDMA policy has contributed to reduced motor-vehicle fatalities but not to a cessation of underage drinking.
A multi-faceted approach is needed: enforcement (drunk driving penalties and ignition locks), education (how to drink responsibly), and cultural shifts (engagement, service, and accountability).
Radical policy considerations (lowering or abolishing the MDMA and universal ignition locks) are presented as controversial possibilities to address deeper structural issues around alcohol culture on campuses.
The author emphasizes the complexity of the problem, the need for actionable strategies beyond prohibition, and the importance of addressing student perceptions and knowledge gaps about safe drinking.
(End of notes)
READING 2:
Drink and drive? Understanding the dynamics of youth risk-taking (Boes & Stillman, Health Economics, 2024)
Purpose and design
Exploit New Zealand’s (NZ) policy change: minimum legal drinking age (MLDA) reduced from 20 to 18 in 1999 to study dynamics of youth risk-taking.
Data: universe of road accidents over 15 years (1996–2011) plus health survey data and police BAC tests.
Two complementary analyses:
Event history approach (time-based regression discontinuity around the policy change) to estimate short-run impacts by age group.
Cohort analysis comparing outcomes across groups exposed to different MLDAs, with graphical and regression components to trace longer-run dynamics.
Main finding: lowering MLDA from 20 to 18 did not increase alcohol-related or total vehicular accidents among teens in NZ in the short run or in cumulative terms for affected cohorts.
Key concepts and context
MLDA: legal threshold for purchasing alcohol; changes may affect youth drinking, driving, and broader risk-taking.
Regression Discontinuity Design (RDD): leverages a discrete change at a threshold (MLDA) to identify causal effects around the threshold.
Two limitations highlighted for age-based RDDs:
They identify short-run (around the threshold) effects but not necessarily persistence across ages.
They may not capture dynamic, cross-age, or cohort-wide adjustments; hence the paper’s focus on dynamic (time- and cohort-based) analyses.
The NZ context includes other policy changes under SLAA1999 (Sale of Liquor Amendment Act 1999): supermarkets could sell beer, Sunday alcohol sales liberalized. Authors note these could confound estimates but argue the main results (no MLDA impact) likely bound the true effect.
Policy change specifics
MLDA lowered from 20 to 18 on December 1, 1999; policy announced August 31, 1999.
The policy change is treated as a natural experiment; enforcement intensity and other regulatory changes may modulate effects.
Public discussion and high visibility of the MLDA change may have induced observational targeting or “observation effects” that influence youth behavior.
NZ policy environment during period included ongoing enforcement and various changes to alcohol access; the authors discuss potential spillovers and learning-by-doing effects.
Data and outcomes
Accident data: NZ Ministry of Transport; monthly/weekly records 1996–2011; include location, time, vehicle type, passengers, driver sex, date of birth, and police judgment of alcohol involvement.
Population denominators: Statistics NZ population estimates by age, gender, and quarter; used to compute rates for the cohort analysis (per 10,000 population).
Outcomes studied (three):
(i) Total number of accidents.
(ii) Number of alcohol-related accidents (positive BAC or police suspicion of alcohol involvement; defined robustly as in coding rule described in endnotes).
(iii) Proportion of accidents that are alcohol-related (relative to total accidents).
Data for behavior and consumption:
NZ Health Survey (NZHS): 1996/97, 2002/03, 2006/07 waves; measures intensive and extensive margins of youth alcohol use.
Police BAC tests (blood alcohol concentration) for accident-involved youths; administrative data for test results ( BAC in mg per 100 mL ).
Additional context: driving license progression in NZ; typical ages to obtain licenses; legal BAC limits during period: ext{BAC}{>20} = 80 rac{ ext{mg}}{100 ext{ mL}}, ext{BAC}{ ext{}
}Note: The manuscript contains figures (Figure 1–8) and tables (Tables 1–3). The notes below summarize their implications without reproducing every graph.
Data preparation and design details
Aggregation: accident data aggregated to weekly levels for event-history analyses; further smoothing with local linear regression in plots.
Age groups analyzed separately:
15–17 years (potential indirect impact via easier access)
18–19 years (directly affected by MLDA change)
20–21 years (not affected by the policy change; used as a potential control group)
Key dates: policy change date December 1, 1999; announcement date August 31, 1999; both incorporated in event-history analyses to handle timing around the change.
The authors comment on secular trends and seasonality, including week-of-year and day-of-week effects; controls included where appropriate.
Event-history (time-based) analysis: core model and interpretation
Goal: identify a causal short-run impact of the MLDA reduction by comparing outcomes just before and just after the policy change, controlling for time trends.
Core specification (Equation 1):
where:
y is one of the three outcomes (total accidents, alcohol-related accidents, share of alcohol-related accidents).
is a post-change indicator (1 if after December 1, 1999; 0 otherwise).
and are flexible functions of time, capturing pre- and post-change time trends, with time normalized to zero at the policy change.
captures the causal impact of MLDA change under the quasi-experimental setup.
Estimation approach:
Nonparametric (local polynomial) in time around the policy change; uses Epanechnikov kernel; bandwidth chosen via rule-of-thumb (Imbens & Kalyanaraman, 2012).
Parametric specification as a robustness check: include pre/post time indicators and time trends.
Data granularity in figures: weekly observations around the policy change; main results summarized over 25 weeks after vs 25 weeks before the change.
Key finding from nonparametric event-history (Figure 4): no immediate, noticeable discontinuity in any outcome at December 1, 1999, across age groups.
Key finding from parametric event-history (Table 2): no short-run increases in total accidents, alcohol-related accidents, or the share of alcohol-related accidents for 15–17 and 18–19 year-olds; some estimates show modestly negative effects in the 30–89 day window; confidence intervals include null effects; after the initial post-change window, effects are near zero or negative for several outcomes.
Robustness notes (Text): results are robust to excluding the announcement period, controlling for driver gender, location, vehicle type; and to redefining the post-reform window; gender-specific results also show no significant impacts.
Implication: the policy change did not generate a detectable short-run increase in accidents or drinking behavior among youths in the immediate months after the change, given the data and model specification.
Cohort analysis: dynamic age-by-cohort patterns (Section 4.2)
Cohorts defined relative to age at policy change (birth-year alignment):
Cohort 20+ : age 20+ at change; unaffected by MLDA change (counterfactual for younger cohorts).
Cohort 18–20 : aged 18–20 at change; directly affected by MLDA change.
Cohort 15–18 : aged 15–18 at change; became legal drinkers on their 18th birthday.
Cohort 15– : younger cohort; became legal drinkers on their 18th birthday, but the change occurred before they’d be driving.
Raw patterns (Figure 5): accident rates (total and alcohol-related) rise with age from 16 to ~18.5, then decline. The share of alcohol-related accidents increases sharply from 16 to ~18.5, then stabilizes around 15–20% through age ~24.
Cumulative accidents (Figure 5 bottom panels):
20+ cohort has higher cumulative accident counts up to around age 21.
18–20 and 15–18 cohorts exhibit notably lower cumulative accidents than 20+ up to ages ~22; differences persist through age 24.
For alcohol-related accidents, the 18–20 cohort shows about 11% fewer cumulative accidents by age 18 and ~13% fewer by age 20 compared with the 20+ cohort; the 15–18 cohort has consistently lower cumulative alcohol-related accidents than the 20+ cohort;
the 15– cohort resembles the 20+ pattern with only small differences.Parametric cohort regression (Table 3): four-year age polynomial in person age, with interactions by cohort to capture cohort-specific age profiles. Key results:
Cohort 18–20: total accidents and alcohol-related accidents lower at younger ages than the 20+ cohort; significant reductions in total accidents; smaller effect on the share of alcohol-related accidents.
Cohort 15–18: increases in total accidents relative to 20+ but smaller magnitudes; alcohol-related accidents lower; share effects modest.
Cohort 15-: patterns similar to 20+ for total accidents and share of alcohol-related accidents; some modest deviations.
F-tests indicate significant differences in age patterns for cohorts 18–20 and 15–18 vs 20+ for total and alcohol-related accidents, but not for the proportion of accidents that are alcohol-related.
Marginal age effects (Figure 6): shows how an extra month of age shifts outcomes. Key observations:
Total accident rates rise with age up to ~18.5, then flatten; pattern weaker (flatter) for the 18–20 cohort.
Alcohol-related accidents show similar dynamics but with pronounced differences at younger ages for affected cohorts.
Birth-year dynamics (Figure 7): across birth cohorts, accident dynamics by age are broadly similar; notable dips in peak accidents for ages 17–19 cohorts, while oldest and youngest cohorts resemble each other.
Interpretation from cohort results:
The MLDA reduction appears not to increase risky driving in the long run; in fact, for some directly affected cohorts, there are lasting reductions in total and alcohol-related accidents up to early adulthood.
The youngest cohort (15–) shows patterns similar to the 20+ cohort, suggesting a fading salience of the policy as time passes.
Impacts on alcohol consumption (Section 5)
Question: does no increase in crashes imply no long-run impact on youth drinking, or is drinking behavior changing without crashes?
Data sources for consumption:
NZ Health Survey (NZHS): waves in 1996/97, 2002/03, 2006/07; measures include:
Probability of drinking in the last year.
Days per month drinking in the last year.
Number of drinks per drinking occasion.
Blood alcohol level (when tested) in accidents.
Police BAC data: blood alcohol concentration for those involved in accidents and tested.
Key findings from NZHS (cohorts 20+ and 15–):
Age patterns of drinking and consumption show striking similarity between the 20+ and 15– cohorts across the same age ranges, despite non-linear trajectories in the non-cohort-specific patterns.
The 15–18 and 18–20 cohorts exhibit slightly different patterns in a few measures, but differences are modest and do not indicate a large long-run shift in youth drinking due to MLDA change.
Frequency of drinking rises with age from about 1 day/month at age 15 to ~4.5 days/month by age 19–20, then stabilizes around 4 days/month; drinks per session rise from ~2 drinks at age 15 to ~5 drinks at age 19, then level around ~4.5 drinks.
Blood alcohol levels (BA) and other consumption indicators:
BA levels are generally lower for the 15–18 and 18–20 cohorts from age 18 onwards, consistent with lower risk exposure or moderation in the long run for these groups.
In the short run (around the policy change), there is evidence of more harmful drinking among teenagers, consistent with a possible immediate response to gaining legal access, but this does not translate into higher long-run risk on the roads.
Overall interpretation on consumption:
The MLDA change in NZ appears uncorrelated with long-run trends in youth risky driving and drinking; short-run increases in risky drinking may occur, but long-run drinking patterns converge across cohorts.
Practical considerations:
Extensive public discussion and heightened observer effects around the change likely contributed to heterogeneous responses: some teens may have reduced drinking when intending to drive, while others may have increased drinking, offsetting effects on accidents.
Additional methodological and interpretive notes
The authors discuss potential explanations for null or small effects on accidents post-change:
Incomplete enforcement of MLDA
Uniform effectiveness or ineffectiveness across age groups
Peer effects and learning-by-doing that attenuate or counterbalance direct effects
They acknowledge generalizability concerns:
NZ’s regulatory environment and enforcement may differ from other countries; results may not transfer to jurisdictions with different alcohol-control regimes.
Comparisons to other studies:
Some studies using age-based RDD (e.g., in NSW, Australia) find increased drinking at the threshold but no clear effects on accidents; the NZ study suggests that such short-run effects do not necessarily translate into long-run changes in risk behavior.
Implications for policy evaluation:
The study highlights the importance of analyzing dynamic, cohort-based outcomes in addition to traditional RDD analyses when assessing age-based policies.
It cautions against extrapolating short-run RDD results to long-run or cross-age impacts without considering dynamic behavioral responses.
Equations and technical details highlighted in the study
Short-run event-history model (causal effect around threshold):
where
if time is after December 1, 1999; 0 otherwise.
is time relative to the policy change (normalized to zero at the policy change).
Cohort regression specification (controls for cohort and age patterns):
where
y is one of: (1) total accidents per 10,000, (2) total alcohol-related accidents per 10,000, (3) proportion of alcohol-related accidents relative to total.
age_j denotes polynomial terms in age (up to degree 4); main effect for the 20+ cohort; interactions with cohort indicators (18–20, 15–18, 15−).
X includes covariates to model seasonality and time trends; models allow for age-specific seasonality and cohort effects; standard errors clustered at the crash-month level.
Data domain and time windows
Event-history analysis uses 548 days before to 183 days after the change in the key hourly regression for the three outcomes (total, alcohol-related, and proportion of alcohol-related accidents).
The cohort analysis tracks monthly data from age 16 to 24; cumulative observations around thresholds reflect long-run dynamics.
Descriptive and robustness checks (highlights)
Descriptive (Table 1): no strong pre/post change differences in 15–17 or 18–19 groups for total or alcohol-related accidents; small indications for 20–21 group with reductions in some measures.
The literature review and robustness checks emphasize that earlier DID approaches (without controlling for age-specific trends and pre-existing differences) could yield biased estimates; the current study argues that event-history and cohort analyses address these concerns.
The authors acknowledge that data noise and potential confounders (e.g., other policy changes around SLAA1999) exist but argue that estimates are upper-bounds for the MLDA effect when holding other changes constant.
Conclusions and policy implications
The NZ MLDA reduction from 20 to 18 in 1999 did not lead to a detectable increase in teen motor-vehicle accidents or alcohol-related injuries in both the short run and the longer run for affected cohorts.
The targeted cohort analysis suggests potential short-run reductions in risky driving among youths who were already 15 at the time of the change, but no lasting adverse effects on younger cohorts.
Implications for policy design: MLDA changes can be potentially implemented without necessarily triggering negative outcomes for youths, though local enforcement, public discussion, and context-specific factors matter.
Caution against relying solely on age-based RDD to infer long-run policy effects; dynamic analyses across ages and cohorts provide a more complete picture of the effects of age-based policies.
Ethical, philosophical, and practical implications
Ethical: balancing public health goals with individual freedom; evaluating whether the transition costs (in terms of short-run shifts in drinking or perceived risk) justify changes in MLDA.
Practical: the public discourse around policy changes can influence behavior beyond the legal threshold; policymakers should consider domain-specific context, enforcement capacity, and communication strategies when implementing age-based policies.
Relevance to other age-based policies: findings stress the importance of dynamic, cohort-aware analysis when evaluating policies like the smoking age, retirement age, or youth minimum wages.
Connections to previous lectures and broader literature
Aligns with a large literature showing MLDA restrictions can reduce harms, but emphasizes that short-run RDD evidence may not extrapolate to longer-run effects.
Reinforces methodological point: RDD near a single threshold is informative about short-run local treatment effects; dynamic policies require time-varying and cohort-oriented analyses to capture persistence and cross-age effects.
Related studies cited include work by Conover & Scrimgeour (2013), Kypri et al. (various years), Lindo et al. (2016), Gupta & Nielsson (2017), and Lazo et al. (contextual comparisons across countries).
Core numerical references and formulas (LaTeX) used in the analysis
Event-history model (central causal equation):
Cohort regression specification (polynomial age terms and cohort interactions):
Blood alcohol limit context (as stated in text):
ext{BAC}{>20} = 80 \frac{\text{mg}}{100\,\text{mL}}, \quad \text{BAC}{\le 20} = 30 \frac{\text{mg}}{100\,\text{mL}}.
Note: All mathematical expressions are presented in LaTeX format as per the guidance; use the double-dollar syntax in documents when rendering.
Takeaways for exam preparation
The NZ MLDA change provides a natural experiment to study not just short-run jumps at the threshold, but dynamic, cross-age, and cohort-level responses.
Event-history (time-based) analyses can reveal the absence of immediate risk increases and help bound policy effects; cohort analyses can reveal longer-run dynamics and potential learning-by-doing or peer effects.
The study illustrates how policy evaluations can yield different conclusions depending on the methodological lens (RDD around a threshold vs. cohort dynamics across ages).
Long-run policies require careful interpretation of consumption and behavior data to distinguish true absence of effect from offsetting mechanisms (e.g., enforcement, public discussion, observation effects).
References cited (selection)
Angrist & Rokkanen (2015); Imbens & Kalyanaraman (2012) on bandwidth selection; major related empirical studies on MLDA impacts across various countries; NZ-specific studies on hospitalizations, drinking contexts, and violence around MLDA changes.
Summary verdict
The evidence suggests that reducing NZ’s MLDA from 20 to 18 did not escalate youth driving risks or alcohol-related harm in the short run or across cohorts in the longer run; minor, context-specific patterns emerge in the short run and across certain birth-year cohorts; long-run drinking patterns show little systematic change across cohorts.
Practical takeaway for exam style questions
If asked to summarize results: state the null findings for short-run and long-run accident outcomes by age group and by cohort, and emphasize the role of dynamic analysis in understanding policy effects beyond local RDD conclusions.
If asked to discuss methods: describe how the event-history model (Equation 1) and the cohort regression model (Equation 2) work, including the interpretation of β and cohort interaction terms, and explain why both approaches complement each other.
If asked to discuss policy implications: articulate that MLDA policy changes can be compatible with stable youth risk profiles in some contexts, but the external validity depends on enforcement, outlet access changes, and public discourse; caution against overgeneralizing RDD findings to long-run effects.
References to figures and tables (for quick recall during revision)
Figure 1: Early NZ mortality patterns by MLDA threshold (1988–1999 vs 2000–2016) showing pre/post dynamics and the 18 threshold change.
Figure 2: Vehicular accidents by age around MLDA change; no clear discontinuities at thresholds.
Figure 3: Development of accidents over time by age group (monthly data; 3 outcomes).
Figure 4: Event-history (weekly) results around policy change; no evident discontinuity.
Table 1: Descriptive statistics by pre/post policy change and age group.
Table 2: Event-history results (short-run) across 0–29 days, 30–89 days, and 90–183 days windows post-change.
Table 3: Regression estimates by cohort for total accidents, alcohol-related accidents, and share of alcohol-related accidents.
Figure 5–7: Cohort accident dynamics, fitted age profiles, and birth-year patterns.
Figure 8: NZHS alcohol consumption patterns by cohort.
Final note
The paper argues that the MLDA can be lowered without necessarily triggering detrimental outcomes for youths, and it highlights the importance of matching policy evaluation methods to the dynamic, multi-age nature of behavioral responses. It also suggests that previous conclusions drawn from age-based RDDs should be interpreted with caution when long-run and cross-age effects are of policy interest.
READING 3:
Introduction
The article examines a well-established decline in youth drinking in most high-income countries over the past two decades and explores its implications for public health, policy-making, and public debate.
Core motivation: While drivers of the decline have been studied, its implications for health and policy have received less attention (
public health importance of youth drinking trends given alcohol as the leading risk factor globally for mortality and morbidity among ages 15-24).Key idea: A decline in youth drinking could yield large short- and long-term public health benefits, but may also complicate population-level harm patterns as heavier-drinking cohorts age into higher-risk periods.
Two illustrative policy models are used to think about potential futures: the reinforcement model and the withdrawal model.
COVID-19 adds uncertainty about post-pandemic trajectories in youth drinking and related harms.
What is known about the decline in youth drinking
Decline observed broadly across Western/high-income countries since the mid-2000s; cross-country variation exists in magnitude and timing (HBSC 2018; ESPAD 2020).
In some countries, the decline appears to be stabilising or slowing but not reversing (HBSC 2018-2019; ESPAD 2018-2019).
Conceptual labels used in public discourse: ‘generation sensible’ and ‘the new puritans’; contrasted with the 1990s-2000s ‘new culture of intoxication’.
Alcohol is a major global health risk for youth and can affect road safety, violence, education, mental health, and later-life alcohol-related disease.
There is concern that declines in youth drinking could shift harm to older age groups as heavier-drinking cohorts age through life stages.
The COVID-19 pandemic raised questions about changes in drinking during lockdowns and whether these shifts persist post-pandemic.
Evidence base for the decline and its durability
The downturn began in the mid-2000s and spread from the US/Northern Europe to Western Europe and Australia in the late 1990s/early 2000s.
More recent data suggest the decline is stabilising in some places, with no clear reversal yet across high-income countries.
The trend is broad and long-term, suggesting structural/cultural shifts rather than a short-lived phenomenon.
Potential drivers (structural and cultural shifts):
Economic insecurity among youths
Influence of internet-based technologies
Changes in parent-child relationships
Immigration from countries with abstemious drinking cultures
Wellness/healthism and health-promoting practices
Related age-behavior changes across adolescence (Twenge et al.) indicate broader shifts in youth behavior, which may hinder rapid resurgences in drinking.
Debate exists about whether declines in youth drinking translate into declines in adult drinking; some cohorts show persistence of lighter drinking into adulthood, but evidence is mixed by country.
US evidence shows some cohorts (late 1970s/early 1980s birth cohorts) report higher adult consumption than surrounding generations despite lower adolescent drinking, illustrating potential cohort-driven shifts in life-course trajectories.
Examples from specific data points:
England: proportion of 16-24 year-olds who drank in the last week fell from in 2002 to in 2019.
England: hospital admissions for alcohol-specific conditions among under-18s declined by from 72.1 to 30.7 per 100,000 between 2006/07–2008/09 and 2017/18–2019/20.
England: alcohol-specific mortality among 30-34 year-olds decreased by between 2001 and 2019.
Australia: hospitalisation rates for 15-34 year-olds for alcohol-attributable conditions broadly stable from 2012 to 2017.
Population subgroups: declines occur across youth but may be smaller among girls, those in lower socioeconomic positions, and heavier drinkers (evidence mixed).
Long latency of some health benefits means peak reductions in alcohol-attributable mortality/morbidity may occur decades later (roughly after declines).
Implications for public health
Short-term benefits for youths: expected reductions in road traffic accidents, violence, alcohol poisoning, and acute alcohol dependence.
Long-term benefits depend on persistence of lighter drinking into adulthood; if persistence fails, overall reductions in alcohol-related harm could be smaller.
Possible countervailing dynamics: heavier-drinking cohorts aging into middle-to-older ages could keep aggregate harm levels elevated for some time.
Health impact nuances:
Latency and competing trends can blur population-level effects on chronic diseases that peak later in life (ages 45-65).
The harm-reduction benefits depend on co-trends in other risk factors (smoking, illicit drugs, obesity). Some risks are rising (obesity, mental health problems) while others decline (smoking, illicit drug use).
Benefits may be moderated by sociodemographic factors; higher risk from heavier drinking is more pronounced in women and disadvantaged groups; the downturn might be smaller in these groups.
If youth drinking declines persist into adulthood, future harm distribution could shift away from younger ages toward middle/older ages, altering public-health priorities and policy debates.
The ecological picture: population-level benefits could be obscured if older-age harms rise as a consequence of historical patterns in heavy drinking cohorts.
Cross-country differences matter for public health planning; LMICs show different patterns (rising youth drinking in some contexts), limiting global generalizability of the optimistic outlook.
Implications for public policy and debate: two model scenarios
The authors present two illustrative pathways to explore how policy might respond to the decline in youth drinking:
Reinforcement model
Withdrawal model
The reinforcement model
Core idea: As fewer people drink and as younger cohorts carry lighter drinking into adulthood, public support for alcohol control policies grows; policy success is perceived, which further strengthens policy actions.
Potential dynamics:
Public health advocates gain greater influence within policy networks; policies seen as effective and beneficial for public health.
Policy-makers are rewarded politically for strong, evidence-based alcohol control measures, similar to tobacco control trajectories.
External pressure from supranational bodies (e.g., EU/WHO) may compel reforms in otherwise resistant governments.
Distinctions from tobacco: there is less evidence that alcohol-control policies directly caused declines in youth drinking; alcohol is not a binary behavior like smoking, and industry partnerships historically framed many alcohol policies.
Challenges for the reinforcement model:
Lack of clear, health-driven, youth-focused policy successes; health concerns about alcohol are less direct than tobacco's.
The continuum nature of alcohol consumption makes consensus around a single anti-drinking stance harder to articulate.
Even with declines, industry arguments (e.g., impacts on light drinkers) can still resonate, hindering radical reforms.
Overall: reinforcement is plausible in some contexts but has important limitations; incremental rather than radical policy changes may be more likely where public-health-led breakthroughs are absent.
The withdrawal model
Core idea: The public becomes abstemious but apathetic toward solving alcohol-related problems; the alcohol industry leverages the decline to reassert partnerships with government and to focus on other issues.
Policy-makers might deprioritize alcohol, viewing it as less urgent compared to other public health challenges, allowing the industry to re-expand influence.
In countries with strong pre-existing alcohol-policy traditions (e.g., Sweden, Finland, Scotland), the withdrawal model could play out differently but still involves greater industry argumentation and reduced restrictiveness.
Primary consequences:
Relaxation of formal and informal policy controls on alcohol use.
Re-emergence of Skog’s long waves of alcohol consumption (prolonged cycles of rising and falling consumption spanning decades).
Increased lobbying by global corporations for partnerships in new jurisdictions; public health actors may redirect focus to other issues or to narrower, higher-profile campaigns (e.g., Dry January, calorie labeling, no- or low-alcohol products).
The withdrawal scenario implies a potential drift away from strict population-level interventions (like pricing or comprehensive advertising controls) toward partial measures with limited scope.
Moderating factors: in some contexts, policy-makers may still address tractable problems or high-profile campaigns, but overall pressure from industry could erode broader health-promotion measures.
Comparative assessment of the two models
The authors view disconnects between the tobacco experience and the reinforcement model as making withdrawal a more plausible scenario in many settings due to:
Weak direct causal links between youth drinking declines and policy changes.
The absence of clear, universal public-health-led drivers for aggressive, population-wide interventions on alcohol.
Nevertheless, both models are illustrative and not predictive; real-world outcomes may reflect a mix of pathways across countries and over time.
Additional considerations for policy debate and practice
Policy environment weaknesses that can persist irrespective of consumption trends:
Tax systems that do not adequately discourage consumption or reflect externalities.
Dysfunctional self-regulation of advertising by industry bodies.
Inadequate provision of specialist alcohol treatment services.
Debates about population-level relationships between average alcohol consumption and harm: evidence suggests that reductions in youth drinking can lower population averages even when harms among older heavy drinkers persist, challenging simplistic population-wide causation claims.
New policy approaches may focus on tractable, less contentious areas (e.g., Dry January campaigns, calories labeling, no- or low-alcohol drinks) rather than sweeping price/availability reforms.
The long-term policy question: to what extent should alcohol policy pursue harm reduction goals given ongoing debates about ultimate aims? The decline in youth drinking could reshape how end goals are framed and pursued.
Four tentative recommendations for public health actors
Continue researching why youth drinking is declining across contexts to assess persistence and how policy measures can reinforce declines as today’s youth reach adulthood.
Understand youth and emerging adults’ attitudes toward alcohol and policy, and track how ideas evolve (e.g., framing of vulnerable or deficient youth versus broader population harms).
Persist in advocating for core weaknesses in alcohol policy environments (tax systems, advertising self-regulation, and treatment capacity) irrespective of consumption trends; adjust messaging where necessary (e.g., clarify causal links between population-level consumption and harm).
Reflect on the end goals of alcohol policy in light of ongoing trends; consider whether reinforcement of restrictive policies is desirable or whether a shift toward more incremental, targeted public-health strategies may be appropriate in different contexts.
Key nuances and methodological notes
The authors acknowledge several limitations:
Their UK/Australian policy perspectives may influence interpretation; other policy contexts may yield different implications.
The analysis focuses on high-income countries; rising youth drinking in some LMICs could contradict the optimistic view.
The paper emphasizes that the two models are illustrative rather than predictive; they aim to provoke debate and analytical thinking about potential futures.
The discussion integrates evidence from multiple sources, including HBSC, ESPAD, national health statistics, and cross-national policy literature.
Data points and numerical references (selected)
Youth drinking decline indicators:
Across multiple high-income countries since the mid-2000s; accentuated in mid-2000s for many places; some stabilization by 2018-2019 data.
England (youth and young adults):
Proportion drinking in last week for ages fell from (2002) to (2019).
Under-18 hospital admissions for alcohol-specific conditions declined by (from 72.1 to 30.7 per 100,000) between 2006/07-2008/09 and 2017/18-2019/20.
Mortality and morbidity trends: alcohol-specific mortality among year-olds fell by between 2001 and 2019.
Australian context: alcohol-attributable hospitalisation rates among year-olds were broadly stable from 2012 to 2017.
Other trend indicators: decline observed in many cohorts; some evidence of cross-cohort convergence in drinking behaviors, though not uniformly across all age groups or countries.
Key references to models and theories (e.g., Skog’s long waves, Room & Livingston total consumption model) are used to frame potential policy trajectories.
Notable concepts, terms, and links to broader literature
Generation sensible and new puritans (public discourse labels).
New culture of intoxication (earlier era contrasting frame).
The total consumption model: distribution of drinking in a population affects overall harm; policy implications for pricing/availability.
Skog’s long waves of alcohol consumption: cycles spanning decades linked to social and cultural processes.
Policy-transfer and supranational influences: how international norms may pressure national policy changes.
Public health framing versus industry framing: tensions in policy debates around alcohol control policies.
Limitations and scope for future work
The analysis centers on high-income countries; applicability to LMICs may differ due to different market structures, alcohol cultures, and policy environments.
The paper underscores the need for ongoing, context-specific research to identify drivers of decline and to monitor how policy contexts evolve as youth drinking trends shift across generations.
References and sources (themes mentioned)
HBSC and ESPAD data and findings on youth drinking trends.
Studies on the health impacts of alcohol (Griswold et al., 2018) and risk factors combining with drinking (Taylor & Rehm, 2006; others).
Policy and public health literature on tobacco control as a potential comparative analogue (Nathanson, 1999; Studlar & Cairney, 2014).
Discussions of industry influence and policy processes (Miller & Harkins, 2010; Pettigrew et al., 2018).
COVID-19-related changes in alcohol use (Acuff et al., 2021).
Key works on policy theory and public policy processes (Cairney, 2012; Dolowitz & Marsh, 2000).
(Note: All numerical references are presented in LaTeX format where appropriate, enclosed in double dollar signs, e.g., .)
READING 4;
Notes on the study: Minimum legal drinking age and long-term alcohol-attributable morbidity and mortality in Finland (Lancet Public Health, 2023)
Background and purpose
MLDA (minimum legal drinking age) as a policy tool to prevent youth drinking and short-term alcohol-attributable harm, with scarce evidence on long-term health effects.
This study exploits the Finnish 1969 reform that lowered MLDA from 21 to 18 and allowed medium-strength beer sales, creating differential exposure by birth cohort.
Research question: Do birth cohorts exposed to a lower MLDA experience higher alcohol-attributable morbidity and mortality later in life, and is the effect moderated by education?
Follow-up: up to 36 years for alcohol-attributable morbidity and mortality from ages 27–63 years.
Context: the Finnish alcohol reform of 1969
Pre-1969 system: off-premises alcohol sales mainly via state monopolies; strong urban–rural differences in availability.
Reform (Keskiolutlaki 462/1968) effective 1969:
MLDA lowered from 21 to 18 years.
Shops could sell medium-strength beer (ABV up to 4.7%).
State monopoly stores and bars could be established in rural areas.
Estimated impact: per-person alcohol consumption rose markedly; on- and off-premises sales increased substantially within ~5 years.
Implication for study design: cohorts born across 1944–1954 reached different MLDA ages in 1969, enabling quasi-experimental comparison of later-life outcomes.
Study design and data sources
Design: retrospective national cohort study using register data with long follow-up.
Population: birth cohorts 1944–1954 identified from the 1970 census and followed for alcohol-attributable outcomes.
Data sources:
Census data (age, sex, region) from 1970; 1975; 1980; and later population registers.
Care Register for Healthcare (inpatient hospital care data; Finland).
Cause-of-Death Register (Statistics Finland).
Follow-up period: from age 27 (earliest available outcome data in 1971) to age 63 (latest data from 2017).
Exclusions: Åland region residents excluded in 1970; emigration censoring applied.
Representativeness: registers cover the Finnish population; data linked at individual level.
Approvals: Statistics Finland Board of Statistical Ethics; Findata approvals; data handling compliant with Finnish and EU data protection regulations.
Exposure and cohort definitions
Exposure variable: age at which MLDA applied to the individual’s birth cohort, determined by birth year:
1944–1947 cohorts reached MLDA at age 21 (pre-reform exposure: 21+).
1948–1950 cohorts reached MLDA at ages 18–20 (partial reform exposure: 18–20).
1951–1954 cohorts reached MLDA at age 18 (full reform exposure: 18).
Reference category: the 1951 birth cohort (aged 17 at reform) used as the reference in hazard models.
Main comparison: cohorts exposed to lower MLDA (18–20 at reform) vs those exposed to higher MLDA (21 at reform).
Outcomes are analyzed separately for men and women.
Outcomes and how they were measured
Alcohol-attributable morbidity:
Inpatient hospital care with alcohol-attributable diagnoses (primary or secondary, depending on data availability).
Includes direct alcohol-attributable conditions (e.g., alcoholic liver disease, alcoholic pancreatitis; alcohol-related mental/behavioral disorders) and other abrupt alcohol-related causes (e.g., poisoning).
Alcohol-attributable mortality:
Deaths where alcohol is the underlying cause, identified from the Cause-of-Death Register using ICD-based alcohol-attributable codes.
Data coverage: registers with onset of alcohol-attributable hospital care starting 1 January 1971; deaths follow-up through 2017.
Additional outcome analyses detailed in appendix (not reproduced here): chronic alcohol-attributable illnesses; sensitivity analyses for competing risks and age-specific subsamples.
Statistical analysis approach
Primary method: Cox proportional hazards regression with attained age as the time scale.
Stratification/control:
Baseline hazards allowed to vary by municipality of residence in 1970 to account for area-level differences in alcohol availability and other regional factors.
Education used as an effect modifier: completion of at least upper secondary education vs no education.
Separate analyses by sex (men and women) due to possible gender differences in drinking and health outcomes.
Outcomes analyzed:
Alcohol-attributable morbidity (hospitalizations)
Alcohol-attributable mortality (deaths)
Additional analyses outlined in the study:
Morbidity and mortality restricted to chronic alcohol diseases (appendix p 2).
All-cause mortality as a sensitivity check.
Competing risks: Fine-Gray subdistribution hazard model treating external causes (accidents, violence, suicide) as competing events for alcohol-attributable mortality.
Age-stratified analyses: 27–49 years and 50–63 years for morbidity and mortality, with special handling for those with prior hospitalizations.
Additional controls: municipality of birth and residence (1975) to account for urbanisation.
Data handling notes:
Endogeneity concerns addressed by using education attainment as a fixed baseline measure (not a post-reform mediator).
All models adjusted for municipal-level baseline hazards; no direct alcohol intake data available.
Key results: mortality and morbidity by age at reform
Overall pattern: older cohorts (those who were 21+ at reform) had lower alcohol-attributable morbidity and mortality than cohorts exposed to 18–20 at reform.
Morbidity (hazard ratios, HRs) for those aged 21 at reform:
Men: HR = 0.89 (95%CI 0.86−0.93)0.89 (95%CI 0.86−0.93) versus 1951 reference cohort (aged 17 at reform).
Women: HR = 0.87 (95%CI 0.81−0.94)0.87 (95%CI 0.81−0.94) versus 1951 reference.
Mortality (HRs) for those aged 21 at reform:
Men: HR = 0.86 (95%CI 0.79−0.93)0.86 (95%CI 0.79−0.93) versus 1951 reference.
Women: HR = 0.78 (95%CI 0.66−0.92)0.78 (95%CI 0.66−0.92) versus 1951 reference.
Later-born cohorts (1952–1954) did not differ from the 1951 cohort in either morbidity or mortality outcomes.
Example by exact age at reform (illustrative):
Men aged 20 at reform, morbidity HR = 0.93 (95%CI 0.90−0.97)0.93 (95%CI 0.90−0.97); mortality HR = 0.83 (95%CI 0.77−0.90)0.83 (95%CI 0.77−0.90) compared with the 1951 cohort (aged 17 at reform).
Education as moderator (gradient by educational attainment):
Adverse health differences were stronger among those with lower education.
Men without secondary education, aged 21 at reform: morbidity HR = 0.85 (95%CI 0.80−0.90)0.85 (95%CI 0.80−0.90) relative to peers in the 1951 cohort with no education.
Men with at least secondary education: morbidity HR = 0.58 (95%CI 0.55−0.62)0.58 (95%CI 0.55−0.62); for 21-year-olds with secondary education, HR = 0.48 (95%CI 0.45−0.51)0.48 (95%CI 0.45−0.51).
Similar patterns observed for women; for mortality, CIs were wide and overlapping for higher education groups.
All-cause mortality showed a general downward trend favoring younger cohorts, suggesting the observed alcohol-attributable harms ran counter to overall mortality trends.
Competing risks analysis (external causes of death as a competing risk) yielded results similar to the main analyses, indicating findings were not driven by non-alcohol-related external deaths.
Age-group differences:
Morbidity differences were more pronounced in ages 27–49 than in 50–63.
Mortality differences were generally consistent across later-life age ranges but with some variation by sex and education.
Sensitivity analyses and robustness checks
Excluded/dealt with chronic alcohol disease-only subsets; results remained similar to main analyses.
Competing risk analyses (Fine-Gray) corroborated the primary hazard results.
Additional analyses limited to age bands (27–49) and (50–63) produced broadly consistent conclusions, with modest amplification of morbidity effects in the younger band.
Sensitivity to the “younger than reform” observation window: analyses including birth cohorts 1955–57 (aged 11–13 at reform) showed no further escalation in alcohol-attributable morbidity or mortality by age 60, suggesting a plateau or limit to the policy’s impact when exposure occurred very late in adolescence.
Robustness checks with urbanisation covariates (municipality of birth and residence in 1975) did not materially alter main results.
Interpretations and implications
Main interpretation: Higher MLDA (i.e., 21) during late adolescence may protect against later-life alcohol-attributable harm, with benefits extending beyond young adulthood.
The observed gradients by age at reform support the idea that late adolescence is a critical period for developing long-term drinking patterns and health consequences.
The effects were strongest among individuals with lower educational attainment, indicating potential widening of socioeconomic health inequalities if MLDA is lowered.
The Finnish context involved concurrent increases in alcohol availability (e.g., beer sale liberalisation, rural outlet expansion) in 1969, so estimates may partly reflect overall availability increases in addition to the age-limit change.
The authors discuss competing explanations for long-term effects, including cumulative exposure due to earlier initiation, cohort-specific drinking cultures, binge drinking patterns, and other unmeasured factors that accompanied the reform.
Mechanisms and plausible pathways
Earlier exposure could lead to longer cumulative opportunities to develop risky drinking patterns, with persistent effects into adulthood.
Late adolescence is a window of brain development and behavioral maturation, where impulsivity, reward sensitivity, and risk-taking could be more susceptible to alcohol’s effects.
Increases in overall alcohol availability during 1969 may have amplified effects across cohorts, not solely through the age limit change.
The association between lower MLDA and higher long-term harm aligns with the idea that restricting access during a sensitive developmental period yields broader health benefits later in life.
Policy relevance and public health implications
Policy takeaway: MLDA is a potentially effective long-term public health intervention, not only for reducing immediate youth harm but also for reducing morbidity and mortality decades later.
Equity considerations: the health benefits of a higher MLDA may be larger for socioeconomically disadvantaged groups; policies should consider how to mitigate potential widening of health disparities.
Context matters: in settings where alcohol availability increases through other channels, disentangling the unique effect of MLDA requires careful quasi-experimental design.
The study supports arguments for maintaining or raising MLDA as part of a comprehensive alcohol policy aimed at reducing long-term health burdens.
Strengths of the study
Long, nationally representative, register-based follow-up with minimal attrition and population-wide coverage.
Exploits a natural experiment (overnight reform in 1969) to observe long-term outcomes (36-year follow-up).
Separate analyses by sex capture gender-specific patterns in alcohol-related harm.
Explicit exploration of effect modification by education reveals socioeconomic gradients in the policy's long-term effects.
Robustness checks including competing risks and sensitivity analyses reinforce the credibility of findings.
Limitations and external validity
Outcome measurement focuses on severe alcohol-related health events (morbidity requiring hospitalization and premature mortality); prevalence of all alcohol-related harm is higher, so results pertain to more severe end of spectrum.
Ecological and quasi-experimental design: individual-level confounding cannot be fully ruled out; no direct measures of individual alcohol intake, age of initiation, or drinking patterns.
Education as a proxy for socioeconomic position is imperfect; residual confounding by other social determinants could bias estimates.
Time-varying co-interventions: other alcohol policies and price changes over follow-up (e.g., 1975 price hikes, 1977–1994 marketing ban, 2004 price cuts) may confound cohort differences; however, the birth-cohort gradient suggests a persistent effect beyond isolated policy shifts.
Generalizability: Nordic context with relatively high per-capita alcohol consumption and binge patterns in the 1970s; results may differ in countries with different drinking cultures and policy histories. Nevertheless, the late-adolescent vulnerability framework could be universal.
Emigration data likely under-reported; censoring could bias results if selective emigration correlated with exposure, though this was not observed to drive main findings.
Connections to prior work and broader literature
Prior studies showed short-term reductions in consumption and youth harm with higher MLDA, and mixed evidence on long-term mortality, especially with regionally varied policies or those reversed quickly.
This study adds to the literature by leveraging a permanent (not reversed) policy change with long follow-up, providing clearer inference about long-term alcohol-related health outcomes.
Comparisons with Swedish regional experiments and US federal changes highlight the importance of study design and timing for detecting long-term effects; this Finnish natural experiment shows gradient effects by age at reform and education level.
Practical takeaways for exam and applied settings
Key finding: Lowering MLDA to 18 years is associated with higher long-term alcohol-attributable morbidity and mortality, particularly among those exposed in late adolescence and among lower-education groups.
The protective effect of higher MLDA appears to extend into mid-adulthood, suggesting policy actions in adolescence can influence population health decades later.
Socioeconomic disparities matter: health benefits of stricter MLDA might be greater for individuals with lower educational attainment, implying equity-focused effects of alcohol policy.
When evaluating policy changes, consider both immediate and long-term outcomes, and account for concurrent changes in alcohol availability and broader sociocultural shifts.
Key numerical references (summary formulas and values)
Hazard ratios (examples):
Morbidity, men, reform at 21 years: HRextmorbidity(extmen,21)=0.89,ext95%CI(0.86,0.93)HRextmorbidity(extmen,21)=0.89,ext95%CI(0.86,0.93)
Morbidity, women, reform at 21 years: HRextmorbidity(extwomen,21)=0.87,ext95%CI(0.81,0.94)HRextmorbidity(extwomen,21)=0.87,ext95%CI(0.81,0.94)
Mortality, men, reform at 21 years: HRextmortality(extmen,21)=0.86,ext95%CI(0.79,0.93)HRextmortality(extmen,21)=0.86,ext95%CI(0.79,0.93)
Mortality, women, reform at 21 years: HRextmortality(extwomen,21)=0.78,ext95%CI(0.66,0.92)HRextmortality(extwomen,21)=0.78,ext95%CI(0.66,0.92)
Mortality and morbidity by age at reform for illustrative cases:
Morbidity, men aged 20 at reform: HR=0.93 (95%CI 0.90−0.97)HR=0.93 (95%CI 0.90−0.97)
Mortality, men aged 20 at reform: HR=0.83 (95%CI 0.77−0.90)HR=0.83 (95%CI 0.77−0.90)
Mortality and morbidity by education (illustrative gradients):
Men without secondary education, aged 21 at reform: morbidity HR=0.85 (0.80−0.90)HR=0.85 (0.80−0.90)
Men with secondary education, overall morbidity: HR=0.58 (0.55−0.62)HR=0.58 (0.55−0.62)
Men aged 21 with secondary education, morbidity: HR=0.48 (0.45−0.51)HR=0.48 (0.45−0.51)
Outcome scales:
Mortality and morbidity measured over ages 27–63 years; all-cause mortality showed a downward trend for younger cohorts, while alcohol-attributable mortality did not follow the same trend.
Summary takeaway
The Finnish 1969 reform provides strong evidence that higher MLDA during late adolescence reduces long-term alcohol-attributable harm, with stronger benefits among those with lower education, underscoring the long reach of adolescence alcohol policy on population health.
READING 5;
Overview
Study by Elizabeth Manton & David Moore analyzing how Australian alcohol policy (national and Victorian) constructs problems related to young adults (ages ) through a poststructuralist lens.
Focus on three problem areas in policy discourse: gender, intoxication, and brain development in relation to binge drinking.
Data sources: five policy documents spanning s (three national policies: 1990, 2001, 2006; two Victorian plans: 2008, 2013).
Methodology: poststructural policy analysis (Bacchi) focusing on what the policy represents as the problem, underlying presuppositions, and silences; two-level content analysis (policy goals/priority areas; full-text thematic analysis) using NVivo10; examined continuities, changes, and silences over time.
Core argument: Australian alcohol policy has largely ignored the over-representation of young adult men in harms, while expanding problem representations around intoxication and brain development; policy has treated intoxication as the leading cause of harm and has mobilised questionable neuroscience to justify age-related policy changes.
Real-world relevance: illustrates how policy can shape what counts as a “problem,” guide or constrain harm-reduction options, and reflect political and industry influences beyond straightforward epidemiology.
Theoretical approach and methodology
Theoretical frame: poststructuralist policy analysis inspired by Foucault; problems are constituted within policy rather than existing outside it ( endogenous problems) (Bacchi, 2009, 2015).
Core analytic tool: Bacchi’s six questions, of which three relevant here are:
What is the problem represented to be in a specific policy?
What presuppositions or assumptions underlie this representation of the ‘problem’?
What is left unproblematic in this problem representation? Where are the silences?
Analytical aims: identify continuities, changes, and silences in policy discourse on young people and the ‘problem’ of alcohol; interrogate supporting research and assumptions.
Document selection rationale:
National policies: National Health Policy on Alcohol (1990); National Alcohol Strategy: A plan for action 2001–2003–04 (2001); National Alcohol Strategy 2006–2011 (extended to 2011).
Victorian policies: Victoria’s Alcohol Action Plan 2008–2013; Reducing the alcohol and drug toll: Victoria’s plan 2013–2017.
Data handling and analysis:
Documents stored and coded in NVivo10; content and thematic analyses conducted.
Two analysis levels: broad goals/priority areas; full-text thematic analysis.
Inter-rater check on themes; focus on changes, silences, and continuities.
Key policy documents and timelines
National policy documents span , , and .
Victorian plans span and .
Time horizon: 25-year window from 1990 to 2013.
Policy documents serve as sites where problems are rendered legible, legible in particular ways, and tied to political interests and evidence bases.
Theme 1: Gender
Historical context (pre-1990): in the 1970s, young adult males were over-represented in alcohol-related harms (e.g., road crashes, assaults). Norms linked masculinity to drinking (e.g., ability to hold liquor).
Policy silence on male harm after 1990: across the five documents, explicit references to male gender are notably absent at the level of broad policy statements and priority actions. Subgroups named include Aboriginal communities, CALD communities, “young people” (often underage), and women/children, but not Young Males.
Evolution over time:
1990 policy omits male–harm relationships, focusing on broader “individuals/people/communities” and subgroups like pregnant women and women/children.
2001 policy highlights higher-risk groups (including young people) but does not foreground young men as a high-risk subgroup in the policy’s strategic emphasis.
2006 policy notes that harms are felt by young adults and mostly by males, yet the recommendations are gender-neutral, with no explicit male-targeted actions.
Victorian plans (2008, 2013) continue to omit explicit gendered analysis in broader policy statements, though mentions of male victims are sometimes contextualized within road crashes or assaults (not as a primary policy target).
Explanations for the silence on male harm:
Possible waning influence of feminist/masculinity critiques on policy post-1990s.
Policy fatigue with gender debates; shift toward focusing on broader categories like ‘youth’ and ‘women and children’.
Emergence of masculinity-focused research on violence, but not consistently integrated into policy.
Growing emphasis on the drinking practices of women and younger subgroups, potentially at the expense of male harms.
Evidence and data considerations:
Epidemiology shows males as higher-risk for certain harms (e.g., road crashes; violent incidents) in various datasets (e.g., Chikritzhs et al., 2003; Begg et al., 2007; Gao et al., 2014; Jiang et al., 2015).
Despite this, the policy discourse largely eschews male-targeted strategies in favor of more diffuse, gender-neutral approaches.
Implications and critiques:
The absence of male-focused attention in policy may misalign with empirical harm patterns and misses opportunities for targeted interventions addressing male risk behaviors.
Raises questions about how policy representations shape resource allocation and prevention priorities.
Calls for integrating gender analysis more consistently in policy to reflect differential harms and social determinants.
Theme 2: Intoxication
Emergence of intoxication as policy priority:
1990: the term ‘intoxication’ is not used; emphasis on alcoholism/dependence and heavy drinking is clearer.
2001: mentions of drinking to intoxication appear in the national strategy background, but the core emphasis remains on patterns of consumption.
2006: intoxication becomes the top policy priority, described as the main driver of alcohol-related harm, with harms felt particularly among young adults. The policy cites that much harm surrounding intoxication is experienced by young adults, and emphasizes intoxication above other conditions like alcoholism or dependence.
Rationale and supporting sources:
The 2006 strategy relies on an international review (Babor et al., 2003) arguing that the main cause of alcohol-related harm is intoxication, listing acute harms (poisoning, injuries) and other outcomes (including suicide, violence).
Measurement and diagnostic framing:
ICD classification changes shape how intoxication is counted:
ICD-9 vs ICD-10 transition reclassifies intoxication as a distinct acute condition (across coding schemes). In ICD-9, intoxication could be subsumed under nondependent alcohol abuse; in ICD-10, it is an explicit acute condition (F10.0: Acute intoxication; F10.1: Harmful use; F10.2: Dependence syndrome; F10.3–F10.9: Alcohol psychoses).
This reclassification elevates intoxication as a measurable, standalone health issue in hospital data.
Victorian policy (2008) uses a broader Mental Health Diagnostic Group category (021: Alcohol intoxication, harmful use, dependence and withdrawal), which bundles acute and chronic conditions and departs from ICD-10 terminology. This grouping can inflate perceived scale of alcohol-related harm.
Implications of ranking and aggregation:
Ranking harms by grouping several ICD categories under a single umbrella (e.g., alcohol-related mental or behavioral problems that include dependence and intoxication) can inflate the priority of intoxication relative to injuries (falls, road crashes) and assaults.
Changes in diagnostic practices and hospital-admission recording (case-mix funding) may further distort trends in intoxication-related hospitalisations, potentially misattributing rises to intoxication rather than to systemic recording changes.
Policy may thus overemphasize intoxication as the leading cause of harm while underemphasizing or misattributing injuries and violence, particularly among young men.
Victorian vs national framing:
2008 plan foregrounds rising intoxication-related hospital admissions in Victorians aged 15–24, linking it to youth drinking patterns; 2013 plan shifts emphasis away from intoxication in favor of broader drug/toll concerns, with less focus on intoxication as a standalone driver.
Critical analysis and implications:
The conceptualization of intoxication as a universal driver of harm rests on contested classifications and data practices; it may obscure the contextual, social, and age/gendered factors that shape drinking contexts and related harms.
Premature prioritization of intoxication risks misdirecting policy resources away from injuries, violence, and other concrete harms, especially among young men.
Calls for careful scrutiny of how data classifications and hospital-admission practices influence policy priorities and the legitimacy of intoxication as the primary harm driver.
Theme 3: Binge drinking and brain development
Early framing of binge drinking:
1990 policy identifies binge drinking as a concern alongside underage drinking and drink-driving; defines binge drinking as “young people consume more than five drinks in a row and reach the stage of being drunk, sick, or passing out.” The main focus remains on habitual heavy drinking rather than neuroscience.
2001 policy shifts emphasis from average consumption to patterns of drinking, including when/where drinking occurs, heavy-drinking occasions, and drinking culture; binge drinking is not explicitly named, but a background paper defines binge as “deliberate drinking to intoxication” and notes it as most common among young people in specific age groups (14–19 and 20–34).
Transition to neuroscience framing (2006 onward):
2006 national strategy explicitly links binge drinking to brain development in young people, arguing that patterns of youth binge drinking are associated with brain damage, and cites the then-prevailing empirical cue to justify raising the minimum legal purchasing age from to .
2008 Victorian plan echoes this concern with a brain-impairment concept (ARBI: alcohol-related brain impairment) suggesting that heavy drinking episodes can cause brain damage even if not chronic, reinforcing a causative link between bingeing and neurodevelopmental harm.
Use of Toumbourou and Hermens et al. (evidence critiques):
The 2006 strategy cites an unpublished submission by health psychologist John Toumbourou (2005) to justify the age increase; the underlying literature relies on Hermens et al. (2013) and other early neuroimaging research, with the claim that neurobiological vulnerability increases risk during adolescence/early adulthood.
2014 Toumbourou et al. article argues that frequent or episodic binge drinking is of particular concern due to neurobiological vulnerability; it cites Hermens et al. (2013) summarizing 20 studies on alcohol use disorders, alcohol abuse, dependence, and binge drinking, and claims emerging evidence linking binge drinking to brain changes.
Critical evaluation of Hermens et al. (2013) and related neuroscience literature:
Hermens et al. review notes a paucity of studies on short-term effects of binge drinking in youth and emphasizes that some brain changes may be reversible; they caution about unclear causal pathways and the limitations of biomarkers for predicting long-term outcomes.
Only eight studies directly addressed binge drinking in youths; definitional heterogeneity across studies makes firm conclusions problematic.
Schweinsburg et al. (2010) caution that observed brain-function differences in adolescent binge drinkers may reflect pre-existing differences rather than causal effects of drinking; longitudinal designs are needed to establish causality.
Toumbourou et al. (2014) and subsequent summaries tend to present “emerging evidence” as stronger than warranted, potentially overstating the certainty of a causal binge-drinking–brain-damage link.
Analytical critique and implications for policy:
The shift to a brain-development narrative rests on selective synthesis of neuroscience, often privileging rapid, decisive policy action (e.g., raising age to 21) over a careful appraisal of methodological limitations and definitional variability (e.g., binge-drinking definitions).
The authors argue that the 2006 and 2008 policy moves were premature, given the limited and heterogeneous evidence base; newer research (Hermens et al., 2013; Toumbourou et al., 2014) has not decisively strengthened the causal link.
Potential consequences include over-medicalizing youth drinking, expanding moral panic around neurodevelopment, and facilitating policy interventions grounded more in neuroscience rhetoric than robust, consistent evidence.
Definitions and definitional issues:
Binge drinking lacks a universally accepted definition across studies, complicating evidence synthesis and the translation into policy.
The notion of a biomarker or irreversible brain damage is overstated when the supporting evidence demonstrates possible reversibility and context-dependence of brain changes.
Overall implications for policy:
Policymaking should resist premature conclusions about brain damage from binge drinking and instead foreground stronger, longitudinal, cohort-based evidence and consistent binge definitions.
The neuroscience narrative can be powerful but must be integrated with rigorous epidemiology and social-contextual analysis to avoid oversimplified policy prescriptions.
Data sources, measurement issues, and the epidemiological context
Epidemiological observations relevant to policy silences and priorities:
Early data (1970s–1980s) highlighted male over-representation in alcohol-related harms; later policy documents reduce explicit gendered framing.
Recent burden-of-disease studies show male over-representation in certain harms, particularly among those in their early twenties, but age- and sex-disaggregated data limitations limit direct policy translation (e.g., Begg et al., 2007; Gao et al., 2014; Jiang et al., 2015).
Data handling and potential measurement biases:
Shifts in ICD coding (ICD-9 to ICD-10) changed how intoxication and related harms are classified and counted (e.g., creation of F10.0 Acute intoxication; F10.1 Harmful use; F10.2 Dependence syndrome).
The Victorian 2008 Handbook used a Mental Health Diagnostic Group (021) that aggregated several alcohol-related conditions, potentially inflating the apparent scale of alcohol-related problems when presented in policy contexts.
Case-mix funding and hospital-admission practices can influence recorded rates of alcohol-related harm, complicating attribution to intoxication or to other conditions.
Livingston (2008) notes that more than half of alcohol-related admissions were due to Acute Intoxication (F10.0) or Dependence (F10.2); however, Livingston’s data show limitations in disaggregation, which policy documents sometimes rely on to claim intoxication as the primary driver of rising admissions.
Future methodological cautions:
An ARC Discovery Project (2014) explicitly sought to determine whether apparent increases in alcohol-related harm are driven by operational or administrative practices rather than actual harm.
The authors caution policymakers to consider changes in data collection, coding, categorization, and health-system practices when interpreting time trends.
Implications for policy and practice
Policy as problematisation:
Alcohol policy functions as a site where problems are constructed; the way problems are framed (gender, intoxication, brain development) shapes which interventions are visible or feasible.
Gender implications:
Ignoring male harms risks under-designing prevention strategies that could reduce road injuries, violence, and other harms, especially among -year-old men.
A more balanced approach would integrate gender analyses that acknowledge differential exposure, risk, and consequences while also recognizing shifts in youth drinking cultures.
Intoxication framing implications:
Prioritizing intoxication as the leading harm can divert attention from other high-harm domains (injuries, violence), particularly those affecting young men.
Diagnostic aggregation and data practices can create an inflated sense of intoxication’s primacy; policy should demand transparent, disaggregated data and critical appraisal of measurement changes.
Brain-development/binge-drinking framing implications:
While neuroscience can illuminate potential vulnerabilities, policy should require robust, longitudinal evidence and clear operational definitions before enacting age-based restrictions or broad public-health campaigns.
The precautionary impulse to raise purchasing age should be weighed against the strength of evidence and alternative policy options (pricing, availability, marketing, venue policies).
Ethical and practical dimensions:
Policy should avoid pathologizing youth or gendered groups without solid evidentiary foundations.
The influence of industry and political economy on policy formation remains a concern (pricing, taxation, late-night trading, industry lobbying).
Public health messaging should reflect nuanced understandings of drinking contexts, cultural norms, and social determinants rather than simplistic intoxication-brain-damage narratives.
Summary of key numerical and definitional references (LaTeX-ready)
Age group of policy focus: years; broader policy categories include “young people” (often under ) and “young adults” (often ).
Policy time frame: spanning (national) and (Victoria) across s.
Quintessential definitions in binge drinking literature:
Binge drinking: >5 drinks on a single occasion.
Brain-development window cited: “before the mid-” (late adolescence to early adulthood).
ICD coding shifts (illustrative codes):
ICD-9: nondependent alcohol abuse; alcohol dependence; alcoholic psychoses; etc.
ICD-10: Acute intoxication; Harmful use; Dependence syndrome; Alcohol psychoses.
Victorian category (2008 Handbook usage): 021 Alcohol intoxication, harmful use, dependence and withdrawal (Mental Health Diagnostic Group).
Data points cited:
Reduction of explicit male focus in policy discourse across documents, despite evidence of male over-representation in harms in epidemiological studies (e.g., road crashes, assaults).
Policy shifts to treat intoxication as the leading cause of harm (2006) and to emphasize neurodevelopmental risk (binge drinking) (2006–2008).
Statement that more than half of alcohol-admissions were for Acute Intoxication (F10.0) or Dependence Syndrome (F10.2) in Livingston (2008) data, with caveats about data aggregation and classification.
References to key sources (for further reading): Babor et al. (2003); Hermens et al. (2013); Schweinsburg et al. (2010); Toumbourou et al. (2005, 2014); Livingston (2008); Chikritzhs et al. (2003); Begg et al. (2007); Gao et al. (2014); Jiang et al. (2015).
Notes for exam-ready takeaways
Policy problematisations often reveal more about political priorities and data practices than about neutral scientific truth; Bacchi’s framework helps uncover what counts as a problem and what is silenced.
In Australian alcohol policy, gender gets progressively de-emphasized in policy statements, despite epidemiological evidence of male-dominated harms, illustrating a disconnect between data and policy framing.
The rise of ‘intoxication’ as a central policy category is as much about classification systems (ICD-10 vs ICD-9) and data aggregation as it is about actual harm patterns; this can create policy incentives to address a category that may not fully capture or explain the harms.
The early neuroscience framing of binge drinking and brain development as a justification for raising the drinking age to 21 is controversial, given methodological limitations and definitional variability in the underlying research; more robust, longitudinal evidence is required for policy to rely on such claims.
READING 6: 2012
Overview
Study title: Examining the Intended and Unintended Impacts of Raising a Minimum Legal Drinking Age on Primary and Secondary Societal Harm and Violence from a Contextual Policy Perspective: A Scoping Review
Authors: Ruud T. J. Roodbeen, Rachel I. Dijkstra, Karen Schelleman-Offermans, Roland Friele, Dike van de Mheen
Journal: International Journal of Environmental Research and Public Health (2021)
Aim: Synthesize both intended and unintended impacts of raising the Minimum Legal Drinking Age (MLDA) using a contextual policy perspective, building on the model by Lanza-Kaduce and Richards [19].
Key finding: 91 studies analyzed; 119 units of information on intended impacts (68% positive) organized into four paths; 43 units on unintended developments (30% positive) across five themes. Also highlight a gap in implementation research.
Implications: Helps assess how well MLDA laws work, informs policy refinement, and underscores the importance of considering unintended developments and implementation processes.
Key Concepts and Definitions
Minimum Legal Drinking Age (MLDA): The legal age at which alcohol can be purchased/consumed. Common example analyzed: raise from 18 to 21.
Intended impact (policy objective): Direct reduction in the availability of alcohol to underage youth, leading to reduced alcohol use and related harms in adolescents and their environment.
Defined in this study as the primary mechanism by which MLDA aims to reduce harm.
Unintended impact: All additional processes, developments, or occurrences in reality caused by raising the MLDA that are not part of the original objective.
Bridging variable: A moderator/mediator measuring drinking patterns in analyses or methodology that precedes or links the MLDA change to outcomes. Used to distinguish primary vs secondary impacts and to identify how analyses account for alcohol use.
Primary societal impact: Direct effects on youth drinking, possession, or purchasing patterns.
Secondary societal harm and violence: Sequential impacts such as traffic crashes, injuries, suicides, births, etc., that follow primary drinking behavior changes.
Paths (as per the extended model):
Path 1: Implementation processes and developments after MLDA change.
Path 2: Primary societal impact (drinking/purchasing patterns).
Path 3: Secondary societal harm and violence without bridging variable.
Path 4: Secondary societal harm and violence with bridging variable.
Five unintended themes: Identified in the literature as part of the broader, unintended developments after MLDA changes (see section 3.3).
Methods (Brief)
Approach: Scoping review following Arksey & O’Malley with Levac et al. enhancements; aims to map breadth and depth of evidence, not to quantify effect sizes.
Data sources: Scientific databases (Web of Science, Sociological Abstracts, PubMed, PsycINFO, Embase) and grey literature (OAIster, GLIN, Opengrey, Google Scholar); expert input from Kettil Bruun Society.
Selection: Iterative inclusion/exclusion focusing on studies addressing raised MLDA; included both quantitative and qualitative studies; emphasized relevance and rigour (burther criteria into realist-like relevance checks).
Data extraction: A single author performed full-text assessment; a second author checked a 10% sample; details extracted included aim, design, policy measures, policy effects, reflections, data sources; some double extraction performed for reliability.
Analysis: Formative thematic content analysis (deductive + inductive) guided by the Lanza-Kaduce & Richards model, extending it to multiple paths and introducing a bridging variable concept.
Note on data handling: When multiple paths or themes appeared in a single study, units of information could be counted across paths; some studies contributed to more than one outcome type.
Results: Overview of the Evidence base
Studies identified and included: 91 studies for extraction after screening (1025 from databases + 799 from grey literature + 46 expert-provided; 13 full-texts not available; 91 final studies used for extraction).
Units of information on intended impacts: 119 across 91 studies; 68.1% positive overall.
Units of information on unintended impacts: 43 across 91 studies; 30.2% positive overall.
Paths identified for intended impacts: four paths (implementation; primary societal impact; secondary societal harm and violence without bridging variable; secondary societal harm and violence with bridging variable).
Bridging variable usage: Present in some secondary harm analyses; used to distinguish direct effects from those mediated by drinking patterns.
Key methodological findings: Only eight studies reported on the implementation process; this is a notable gap given the importance of implementation in policy effectiveness.
Detailed results by path
First Path: Implementation
Included studies: 8 (eight units of information).
Geographic distribution: United States (5), Netherlands (2), Canada (1).
Methods: Mostly survey research; one used mystery-shopping; one used statistical analysis of databases.
Findings:
Strengthening False ID laws associated with a significant 7.3% reduction in younger-than-21 drivers involved in fatal crashes with positive BAC ($7.3 ext{%}$).
After MLDA increase from 16 to 18, mean compliance by 15-year-olds observed to increase (mystery shopper method).
Perceived parental approval of underage drinking decreased post-MLDA increase, with alignment to drinking status rather than age.
Some groups (high school principals, prevention workers, addiction-care staff) perceived no changes in underage drinking or illicit drug use post-change.
Enforcement intensity reported as low, sporadic, and variable due to personnel and competing priorities.
Second Path: Primary Societal Impact (drinking/purchasing patterns)
Included studies: 35 (37 units of information on primary impact).
Geographic distribution: United States (29), Canada (3), Belgium (1), Netherlands (1), multiple European countries (1).
Methods: Surveys (n=27), analyses of existing databases (n=7), qualitative survey (n=1).
Overall findings: 29/35 studies reported significant/relevant impacts on primary drinking patterns.
Direction of effects:
14/29 showed negative (protective) impact on short-term drinking measures (e.g., past month use, number of drinking occasions, binge drinking).
2/29 showed negative long-term reductions in drinking patterns (long-term follow-up).
5 studies reported changes in purchasing behavior by teens (less purchasing in outlets; more obtaining alcohol via parties or others).
8 studies reported decreases in aggregate or per-capita alcohol sales.
Non-significant findings: 8/35 found no significant or relevant impacts on drinking patterns; possible explanations include measurement limitations, confounding by other policies, and changes in enforcement or context.
Moderating factors noted: interaction with other policies (zero tolerance laws, taxes, campaigns) and time-varying implementation differences.
Third Path: Secondary Societal Harm and Violence without Bridging Variable
Included studies: 48 (48 units of information).
Geographic distribution: United States (47), Canada (1).
Data source/type: All used statistical analyses of existing databases (e.g., FARS for crashes).
Findings: 35/48 (72.9%) found significant/relevant impacts (generally decreases in harms like traffic outcomes); 13/48 (27.1%) found no significant impact.
Traffic-specific findings: 39 studies on traffic outcomes; 30/39 (76.9%) found significant reductions in traffic crashes with MLDA increase; breakdown includes reductions in fatalities (19 studies) and broader crash data (11 studies).
Interaction with other policies: Some studies observed joint impacts with beer taxes, seatbelt laws, dram-shop laws; various contexts showed reductions in fatalities when combined with other controls.
Broader harms: Some studies found reductions in non-traffic harms (e.g., teen suicide, sexually transmitted diseases, birth outcomes) when not conditioned on drinking patterns; others found no significant effects on such outcomes.
Reasons for non-significant findings: measurement limitations, secular trends, or other contextual factors influencing outcomes; difficulties isolating the MLDA effect in broader policy environments.
Fourth Path: Secondary Societal Harm and Violence with Bridging Variable
Included studies: 26 (26 units of information).
Geographic distribution: United States (24), Canada (1), Belgium (1).
Data sources/methods: 20 used statistical analyses of databases; 6 used surveys.
Bridging variables used included: self-reported/driving behavior, BAC measures, hospital-based health service usage linked to alcohol, and mediating/controlling for drink-driving at population levels.
Findings: 18/26 (69.2%) found significant/relevant impacts; 7 observed decreases in traffic fatalities (and some reported joint effects with other policies); 11 involved varied outcomes across different subjects (not uniformly positive or negative).
Notable outcomes: reductions in drink-driving behavior; some studies linked MLDA with prenatal drinking outcomes; some observed effects on high school dropout or disorder prevalence; some noted broader societal shifts.
No-significant findings: 8/26 found no significant secondary harm/violence impacts when drinking patterns were accounted for explicitly; reasons included measurement issues, residual confounding, and contextual factors.
Unintended Developments and Implications (3.3)
Scope: 37 studies; 43 units of information; 33 US, 2 Netherlands, 1 Canada, 1 Belgium.
Themes (five identified):
3.3.1 Comprehensive Impact on Adolescents and Commodities
MLDA changes affect not only the target group but younger and older ages as well, suggesting spillover protective effects for 18-year-olds and broader social norms against alcohol use.
Some evidence that MLDA increases create a climate disapproval of alcohol and some drugs; protective effects extend beyond adolescence to other risk behaviors.
3.3.2 Limited Impact on Excessive Elements and problem drinkers
Heavy/problem drinkers’ drink-driving may be less responsive to MLDA changes than moderate drinkers.
Time trends in injuries among adolescents may not show clear reductions; long-term effects on binge drinking among college students are not consistently observed.
A small incremental increase in MLDA (e.g., 18 to 19) may yield smaller impacts than a more abrupt 18 to 21 change.
3.3.3 Substitution of Behaviour (Change in Patterns)
Substitution effects observed: some youth may shift to other substances (e.g., marijuana) or alter where/how they obtain alcohol (home vs. bars, private settings).
Border hopping: travel to jurisdictions with lower MLDA to drink; effectiveness of MLDA increases declines farther from borders; some reductions in fatalities near borders.
Increases in arrests for alcohol-related offenses among under-20s observed in some contexts.
Dutch data show private property drinking by 16–17-year-olds increased after MLDA rise; private/underground consumption patterns shift.
3.3.4 Interdependence with other policy changes
Price/tax effects: price sensitivity for youth drinking can be altered after MLDA changes; higher taxes may have reduced impact on youth fatalities post-MLDA changes in some contexts.
Synergistic effects: combined measures (MLDA increase with higher taxes) can yield larger reductions in accidents; some evidence of cross-age effects (affecting 15–17 and 21–24 year groups).
Potential downsides: higher taxes may increase demand for illegally produced beer; punitive exposure for responsible drinkers in some contexts.
3.3.5 Policy Endogeneity and Reverse Causality
Compliance trends pre-date MLDA changes; some improvements in compliance occurred before MLDA was raised, suggesting pre-existing shifts in social norms.
Early adopter states that implemented MLDA changes without federal mandate sometimes saw larger reductions in youth fatalities, indicating local support/enforcement drives effectiveness.
Ignoring endogeneity may underestimate MLDA impacts; reverse causality (regions with higher lifetime drunkenness more likely to raise MLDA) is discussed as a potential explanation in some studies.
Discussion and Implications
Main takeaways:
Historically, MLDA research emphasized intended effects; this review shows substantial and nuanced unintended consequences and the importance of considering contextual factors.
The four-path model (plus five bridging-variable-informed unintended themes) provides a framework for evaluating policy effectiveness beyond straightforward output reductions.
Implementation research is underrepresented; only eight studies addressed implementation directly, despite its critical role in translating policy into real-world effects.
The literature largely centers on the United States; results may differ in other contexts (e.g., Europe) due to cultural, regulatory, and enforcement differences.
Practical implications for policymakers and researchers:
Plan for border-hopping risks when nearby regions have lower MLDA; coordinate enforcement across borders.
Consider interactions with alcohol taxation, price policies, and other DUI/road-safety measures to maximize joint benefits.
Invest in implementation science: document enforcement intensity, seller compliance, and social-context changes post-MLDA change to understand real-world functioning of the policy.
Include bridging-variable analyses to better capture how changes in drinking patterns mediate broader harms.
Ethical and societal considerations:
MLDA policies raise questions about autonomy, public safety, and social norms; unintended consequences (e.g., substitution to other drugs or private drinking) require careful ethical consideration and targeted prevention efforts.
Strengths and Limitations (as discussed by authors)
Strengths:
Theory-driven, comprehensive synthesis of both intended and unintended impacts with an extended model to accommodate multiple pathways.
Systematic search across databases and grey literature; inclusion of expert input.
Use of bridging variables to distinguish pathways and to account for drinking patterns in analyses.
Limitations:
Possible publication bias; although the study included qualitative data and broader observations, most results are drawn from published studies.
Language limitations and potential omission of native-language results in local contexts.
Predominance of United States data; cross-context generalizability may be limited.
Focus was on MLDA increases; fewer insights on effects of lowering MLDA were covered, suggesting a need for future work in that area.
Conclusions (Key Takeaways)
The scoping review provides a novel, empirically grounded overview of both intended and unintended impacts of MLDA changes.
It emphasizes that legislation does not operate in a vacuum; implementation processes, social context, and interactions with other policies shape outcomes.
Five main implications for future work:
Broaden geographic contexts beyond the U.S. to understand contextual differences.
Strengthen and systematically study implementation and enforcement processes.
Incorporate multiple pathways (primary and secondary outcomes) and bridging variables in evaluations.
Investigate border-related dynamics and policy endogeneity more robustly.
Consider both raising and lowering MLDA in future research to fully map policy effects.
Supplementary Materials (as per the article)
File S1: Final scientific search strategy
File S2: Final grey search strategy
File S3: Invitation sent to experts
File S4: Exclusion criteria used during selection of studies
File S5: Extraction data
Notable numerical references to remember (selected examples)
Total included studies for extraction:
Total units of information on intended impacts:
Proportion of units of information on intended impacts that were positive: 68.1 ext{%}
Total units of information on unintended impacts:
Proportion of unintended units that were positive: 30.2 ext{%}
Number of paths for intended impacts:
Unintended developments themes: themes
Example effect from implementation: a 7.3 ext{%} reduction in younger-than-21 drivers in fatal crashes with positive BAC where False ID laws strengthened [46]
Example finding on compliance: after MLDA increase from 16 to 18, 15-year-olds’ purchase attempts with mystery shoppers showed increased compliance [47]
MLDA change example discussed: raising the MLDA to 21 in Florida (1985) as the basis for the conceptual model viewing unintended and intended impacts [19]
Connections to broader literature and concepts
Contextual policy perspective: Highlights the need to understand how local social context, enforcement capacity, and related policies influence MLDA effectiveness, aligning with responsive and realism evaluation perspectives.
Policy mix and endogeneity: The interaction between MLDA changes and other policies (taxes, DUI penalties, seatbelt laws) can amplify or dampen effects; policy endogeneity can bias observed impacts if not accounted for.
Real-world applicability: Findings offer starting points for policymakers to anticipate unintended consequences (e.g., border hopping, substitution, and broader societal effects) and to calibrate enforcement and complementary policies accordingly.
Ethical implications: The potential for unintended harms or increased penalties for compliant drinkers in some contexts requires careful ethical consideration and near-term mitigation strategies.
READING 7:
Historical and international context of the 21-year minimum drinking age
After the 1933 end of Prohibition, U.S. states could set their own minimum drinking ages. The common choice was years (in states), followed by (in states), (in states), and (in Ohio) [2].
The age of majority often aligned with , the age at which one could vote in state elections in many states [3].
During the Vietnam War era, public pressure argued that a man conscripted to fight should be allowed a beer; many states lowered the drinking age to . The 26th Amendment of granted voting rights to year-olds, effectively lowering adulthood to for voting purposes.
In , the Reagan administration passed the National Minimum Drinking Age Act, raising the drinking age to nationwide by threatening infrastructure cuts to noncompliant states. By , under- access to alcohol was banned across all states and territories, including Ohio.
The strongest advocacy for the raised drinking age came from Mothers Against Drunk Driving (MADD), a nonprofit founded by mothers of victims of alcohol-fuelled crashes; MADD claims the legislation has saved over lives since from traffic-related morbidity [6].
Subsequent evidence has broadened the benefits of a -year drinking age beyond road safety to reductions in alcohol dependence, alcohol-related violence, suicide, and risky sexual behaviours among youth [7]–[10]. Neurodevelopmental data further support the case by showing that alcohol exposure during adolescence impairs neuronal maturation in under-s [11].
International perspective: six other nations maintain a -year minimum drinking age: Sri Lanka, Indonesia, Kazakhstan, Oman, Pakistan, and Palau. Japan is the closest OECD country with a drinking age of . Most OECD nations set the age at ; some, like Switzerland, Belgium, Austria, Germany, and the Netherlands, have an age of .
Australia saw growing public support for a -year minimum by (50.2% in support) versus (40.7%) [13]. The Medical Journal of Australia article by Toumbourou et al. framed a strong clinical and epidemiological case for a -year threshold, gaining substantial media attention and bolstering advocacy by the National Alliance for Action on Alcohol and the Australian Medical Association [1].
The core question remains: what age is the “magic number”? Is a single age simplistic? The article frames three key proof points and explores an alternative license-based model as a thought experiment.
The three conditions for age-21 legislation in Australia: overview
The burden of proof for considering a change to a -year drinking age rests on three conditions:
(i) Alcohol consumption at years causes significant negative outcomes.
(ii) Age-21 regulations reduce alcohol intake among under-s.
(iii) The public health benefit of restricting alcohol for under- outweighs the value of autonomy granted at for voting, driving, and military service.
The question is whether under-s are disproportionately vulnerable to alcohol-related harms to justify restricting autonomy. The United States is unique among Western countries in setting the MLDA at ; the justification historically came from a road-safety lobby in the , but a broad evidence base has since accumulated.
The essay acknowledges that the Australian context requires weighing theoretical public health goals against practical realities, and it proposes an exploration of an individualized licensing model as a thought experiment.
Condition 1 – An age of vulnerability
Neurodevelopmental evidence
Cross-sectional studies link adolescent alcohol use to short- and long-term cognitive impairment, including deficits in information processing, memory, attention, and executive function, with particular risk from binge drinking [11,14,15].
Structural neuroimaging and post-mortem data show impaired white matter development in the prefrontal cortex and fronto-striatal circuits [16]–[18].
Critics argue these neurobiological differences may reflect pre-existing traits that predispose to experimentation, rather than alcohol effects. Clark et al. suggest studies have not adequately controlled for confounding variables; longitudinal data tracking baseline neurobiology before first exposure are needed [19].
Road safety
A 2001 meta-analysis across nine population studies found that raising the MLDA from to reduced overall road-related mortality by 12 ext{ extendash} (( ext{12%})). This aligns with data from MADD and the National Highway Safety Administration.
The question remains whether the protection is age-specific; would raising it to reduce death rates among year-olds similarly? The literature lacks rigorous age-specific evidence.
Australian context features age-dependent driving limits (L- and P- licenses with near-zero BAC tolerance vs full licenses with BAC).
Risk behaviours
Schoolies (Australian end-of-school celebrations): about of attendees drank more than drinks in a single occasion; had risky sexual behaviours [22].
U.S. data (adolescents 12{--}21) show a strong correlation between alcohol excess and physical violence [23], and early consumption (especially binge drinking) may predict later illicit drug use [24].
Associations with suicidal behaviours exist, but causal direction is not fully characterized [25].
Collectively, these findings depict known harms and support a link between policy and outcomes.
Summary: Condition 1 argues that the group is especially vulnerable to alcohol harms on neurodevelopment, road safety, and risk behaviours, providing a prima facie case for targeted policy.
Condition 2 – The power of the law
If the vulnerability premise is accepted, the next key burden is whether a higher legal drinking age translates into real reductions in early-age alcohol use and harms.
Critics warn that stricter laws can push drinking underground and promote unsafe patterns; policy could backfire if it merely delays consumption without reducing overall risk.
International evidence suggests that restrictions do influence behaviours beyond the immediate age bracket:
European School Survey on Alcohol and Other Drugs (ESPAD) shows cross-country differences:
Past-30-day drinking among 10th-graders: % in the US vs % in Denmark, % in Germany, % in France [26].
Proportion intoxicated before age : % (US) vs % (Denmark), % (Germany), % (France) [26].
New Zealand data show that youth well below the legal drinking age access products through older friends and siblings—a “trickle-down” effect [27].
Overall, evidence suggests that upward shifts in the MLDA reduce drinking among s and also reduce initiation among s, indicating a transmission of policy effects to younger cohorts.
Conclusion for Condition 2: Legal restrictions appear to translate into community practice and can reduce early-age exposure, supporting the case for a higher MLDA in a country like Australia.
Condition 3 – A balancing act
The central challenge is whether the negative impact of alcohol on under-s outweighs their autonomy as legal adults (voting, driving, military service).
The rite-of-passage concept: alcohol is culturally tied to adulthood, and raising the MLDA could be seen as paternalistic or overly intrusive.
Reactance concerns: some evidence suggests a reaction where restrictions trigger increased drinking among under-s, but a large meta-analysis by Wagenaar and Toomey argues an inverse relationship between MLDA and overall alcohol consumption across studies from , indicating higher MLDA reduces consumption [29].
Philosophical question: is blanket regulation fair given that harms are driven by a minority of heavy drinkers? The answer is not purely scientific and involves political philosophy about state protection versus personal autonomy.
Practical policy implication: public opinion should shape policy because a one-size-fits-all approach may not accommodate individual maturity differences.
The policy design question: could a more individualized approach better balance autonomy and safety?
An individualized system: a thought experiment
Proposal: individual alcohol licenses.
An -year-old would need to pass a written exam comparable to Responsible Service of Alcohol training.
A point system could track alcohol-related behaviour; licenses could be revoked for offences.
A provisional license system (akin to L- and P-plates) could restrict the type and quantity of alcohol purchasable by youth.
Implementing such a system would require substantial bureaucratic infrastructure but could provide a smoother, graduated transition from adolescence to responsible use.
Conclusions and take-home messages
Toumbourou et al. call for a multi-level advocacy campaign to consider a -year MLDA; the value may lie in initiating dialogue and raising awareness rather than simply achieving a policy outcome.
The author argues that while the age is largely arbitrary and does not perfectly align with individual maturity, the central aim should be to promote mature, sensible approaches to alcohol among adolescents.
A high-profile legislative debate can stimulate policy discussion and education, regardless of the final legal outcome.
Key figures, concepts, and terms to remember
MLDA = Minimum Legal Drinking Age; in the US, established at by law in .
ESPAD = European School Survey on Alcohol and Other Drugs. Key cross-national data on youth drinking and intoxication patterns.
Reactance theory = behavioral response to perceived threat to autonomy; evidence regarding MLDA and drinking behavior is mixed.
Trickle-down effect = early exposure to alcohol among younger adolescents occurs via older peers and siblings when the legal purchasing age is lower.
Responsible Service of Alcohol (RSA) = training standard referenced in licensing-based thought experiments.
Provisional/L- and P- plates = graduated licensing models used as analogies for gradual introduction to alcohol purchase and consumption.
Key data points to recall (in brief):
US MLDA raised to by ; MADD claims over lives saved since then [6].
Road-safety impact: about % reduction in mortality when MLDA rose from to [20].
Public support in Australia: % in vs % in [13].
Cross-national youth drinking data (ESPAD): US ~% (past 30 days) vs Denmark/Germany/France ~% etc.; intoxication before age : US % vs Denmark % [26].
Important caveats: causal direction in neurodevelopmental studies is debated; longitudinal data are needed to establish causality; cultural and societal factors confound cross-country comparisons.
References and further reading (key sources mentioned in the transcript)
Toumbourou J, Jones S, Hickie B. Should the legal age for alcohol purchase be raised to ? Med J Aust, .
Miron J, Tetelbaum E. Does the minimum legal drinking age save lives? Economic Inquiry, .
Law Reform Commission. The law relating to age of majority, etc., Working Paper No. , .
Wagenaar AC, Toomey TL. Effects of minimum drinking age laws: review and analyses of the literature from . J Stud Alcohol Suppl, .
Hingson RW, Heeren T, Zakocs R. Age of drinking onset and involvement in physical fights after drinking. Pediatrics, .
Hingson RW, Heeren T, Winter MR, Wechsler H. Early age of first drunkenness as a factor in college students’ unplanned and unprotected sex attributable to drinking. Pediatrics, .
Squeglia L, Jacobus J, Tapert SF. The Influence of Substance Use on Adolescent Brain Development. Clin EEG Neurosci, .
Hermens D, Lagopoulos J, Tobias-Webb J. Pathways to alcohol-induced brain impairment in young people: a review. Cortex, .
Carlen PL, Fornazzari L, Bennett J, Wilkinson DA. Computerized tomographic scan assessment of alcoholic brain damage and potential reversibility. Alcohol Clin Exp Res, .
National Drug Strategy Household Survey (AIHW) .
Additional sources cited in the text include ESPAD reports, MADD data, NHTSA statistics, and various peer-reviewed articles cited in the reference list.
REDAING 8: 2014
Age-21 Laws: Evidence and Policy Options
Introduction
Harmful alcohol consumption among youth is a prevention priority in Australia. Frequent or episodic binge drinking (consuming 5+5+ standard drinks on a single occasion) is of particular concern due to neurobiological vulnerability during youth.
There is growing evidence that key aspects of brain and related neurocognitive development continue into early adulthood, and that short- and longer-term cognitive impairment during postpubertal and early adulthood is associated with earlier onset of harmful alcohol use. Neuropsychological and brain-imaging evidence links binge drinking or persistent high alcohol use to adverse effects on brain development, notably in the frontal lobe and frontal–striatal circuits, in young people. Findings are interpreted within a developmental framework seeking pathways to alcohol-induced brain impairment.
A pathway-based approach suggests potential benefits from earlier modification of excessive alcohol use patterns; delaying exposure to toxic effects may benefit those with neurodevelopmental delays.
There is a need to introduce effective alcohol control policies targeting youth due to increases in alcohol-attributable hospitalisations and emergency department attendances, and the normalisation of harmful alcohol behaviour in high-profile youth rituals (e.g., schoolies).
A 2010 questionnaire survey of 260 youth aged 17–19 on the Queensland Gold Coast reported:
Most played drinking games: 74.8extextperthousand74.8extextperthousand
Consumed more than 1010 drinks per night: 64.1extextperthousand64.1extextperthousand
Had sex without protection: 18.3extextperthousand18.3extextperthousand
Had sex with multiple partners: 13.9extextperthousand13.9extextperthousand
Advocacy for action has focused on a comprehensive approach (tax reform, increased industry regulation); the alcohol industry resists via product design, advertising, and promotions targeting youth. Public focus has grown through organisations such as the Australian Medical Association and the National Alliance for Action on Alcohol.
The authors argue for extending prevention efforts to include increasing the minimum purchasing age for alcohol from 1818 to 2121 years (age-21 laws).
The minimum purchasing age in Australia is established primarily through state/territory legislation regulating when a licensed venue can sell or allow use of alcohol (e.g., the South Australian Liquor Licensing Act 1997).
Evidence on the impact of age restrictions on youth alcohol harm
Across the US, Canada, New Zealand, and Australia, increasing the legal purchasing age is associated with reduced youth alcohol harm.
Lowering the age to 1818 (to 18 years) has been linked to higher youth harm:
In the US, 29 states lowered the legal drinking age from 2121 to 1818 between 1970 and 1975.
In Canada, all 10 provinces lowered the minimum age to 18; in South Australia, Western Australia, and Queensland the minimum age was also lowered to 18.
A meta-analysis found that lowering the age increased the incidence of crashes involving drivers aged 18–2018–20 by 10extextperthousand10extextperthousand.
A trickle-down effect was observed where harms increased among 15–1715–17-year-olds in some cases.
Conversely, increasing the legal drinking age to 21 reduces youth alcohol-related harm:
In the late 1970s1970s and early 1980s1980s, several US states raised the legal drinking age to 2121 and observed reductions in alcohol-involved traffic crashes.
In 1984, the US Government passed legislation tying highway funds to states enacting age-21 laws; by 19881988 all states complied.
A systematic review of 1717 studies found consistent effects and estimated an average reduction of around 16 ext{%} in underage crash involvement.
Benefits extend beyond adolescence; improved road safety between ages 2121 and 2525 years have been attributed to delayed exposure and subsequent moderation in drinking patterns, as well as stronger enforcement.
In Canada, higher minimum legal purchasing age reduced youth hospitalisation rates for alcohol-use disorders, alcohol poisoning, suicidal behaviour, and traffic crash injury.
Overall conclusion: the evidence strongly suggests that raising the minimum purchasing age for alcohol would reduce youth alcohol-related harm in Australia.
Additional notes:
A box in the article outlines policy options for implementing age-21 laws (see below).
Evidence also discusses the long-term follow-on benefits of delaying alcohol exposure, and how stricter enforcement amplifies harms reduction.
Developmental framework and contextual factors
A developmental framework emphasises ongoing brain maturation into early adulthood and how early exposure to alcohol may disrupt frontal–striatal circuits involved in impulse control and decision-making.
The vulnerability is greatest among those with neurodevelopmental delays or other risk factors.
The social environment, access to alcohol, and enforcement intensity influence youth consumption patterns and harm trajectories.
Public health implications extend beyond individual risk to societal costs (healthcare, policing, lost productivity).
Policy options and implementation considerations (Box content)
The Box presents options for introducing age-21 laws:
Federal-level coordination: A national agreement to amend all relevant regulations to raise the legal purchasing age to 2121 years (analogous to the US approach in 19841984, which tied highway funding to state compliance).
State/territory restrictions: Implement restrictions such as preserving prohibitions until 1919 or 2020 years of age; this removes legal purchasing from school-age populations and can be extended nationally.
Product restrictions: Limit the amount and types of alcoholic products accessible to youths (as practiced in Norway and Sweden).
Secondary supply controls: Several jurisdictions (Northern Territory, New South Wales, Queensland, Tasmania, Victoria) have limited secondary supply to minors; extending this nationally could be part of the strategy.
Context-specific use restrictions: Limit alcohol use in public spaces and other contexts.
Additional policy design elements: enforcement intensity, public education campaigns, and targeted communication to counter vested interests.
How the policy would operate in practice (policy options and strategic considerations)
A politically challenging approach could start with a federal brokered agreement to raise the purchasing age across states/territories (with federal incentives or conditions).
Less challenging steps could include state-level actions to raise the age or restrict access, while progressively expanding restrictions to other jurisdictions.
Enforcement, compliance checks, and penalties are critical to achieving meaningful reductions in youth harm.
Complementary measures include public health messaging about youth vulnerability and evidence on benefits, and ongoing research to monitor impact.
Four objections and responses often raised against age-21 laws
1) Autonomy and fairness: “If you are old enough to go to war, you should be old enough to drink.”
Response: Young people are neurologically not fully mature at 1818; there is increasing vulnerability to alcohol harm. Lowering the purchasing age has been associated with at least a 10 ext{%} yearly increase in youth harms, suggesting the policy could prevent more harm than contemporary conflicts.
There is a need to protect young people from harm (including second-hand effects) and to recognise their rights to protective policies.
2) Public support among youth: age-21 laws may lack broad support and could alienate younger voters.Evidence of public opinion trends shows growing support: from 40.7 ext{%} in 2004 to 50.2 ext{%} in 2010, indicating a mounting consensus that such laws are acceptable or desirable. Involving youth in discussion may increase awareness and acceptance.
3) Drug substitution risk: would raising the drinking age push youths toward other drugs?Cross-national data do not support this concern. In the US, after age-21 laws were introduced, alcohol use declined without a compensatory rise in illicit drug use. A longitudinal follow-up (2010–2011) showed that after the age threshold, alcohol use remained lower in the US, while illicit drug use rates were similar to prior levels.
Cross-national comparison (2002) showed higher abstinence from alcohol, tobacco, or illicit drugs among US youth (69%) compared with Australia (42%).
4) Relevance given targeted road safety initiatives: New Zealand’s experience suggests that reducing the minimum purchasing age can still increase injuries among older youth if not carefully implemented.In NZ (1999), lowering the age from 20 to 18 led to more injuries among 15–19-year-olds due to broader market changes; this supports caution and the need for a comprehensive, well-enforced approach rather than isolated changes.
Advocacy and knowledge translation strategies
Four-step plan for effectively advocating age-21 laws in Australia:
Step 1: Build endorsements from public health, law enforcement, and other professional and citizen organisations; develop a coordinated national, state, and territory advocacy program.
Step 2: Maintain public focus on the vulnerability of youth and the likely benefits of age-21 legislation.
Step 3: Provide politicians with briefing material and ready responses to concerns from the community and the alcohol industry.
Step 4: Sustain advocacy through phased, multi-jurisdictional adoption, recognising that opportunities may arise in one jurisdiction first and others may follow.
Knowledge translation literature suggests disseminating research to key political constituencies and countering vested interests with evidence-based messaging.
Author information, conflicts of interest, and provenance (as disclosed in the article)
Authors and affiliations:
John W. Toumbourou, PhD; Professor and Chair in Health Psychology, Deakin University
Kypros Kypri, PhD; Professor and Senior Bell Fellow, University affiliations
Sandra C. Jones, MBA, MPH, PhD; Professor and Director, University affiliations
Ian B. Hickie, MB BS, MD, FRANZCP; Professor and Executive Director, Brain & Mind Research Institute
Acknowledgement: Thanks to Mike Daube for advice in drafting the manuscript.
Competing interests: Statements of funding and affiliations for each author (NHMRC, ARC, Health Research Council of NZ, DrinkWise, NSW Health, etc.).
Provenance: Not commissioned; externally peer reviewed.
Notes on references and related literature (contextual)
The article cites multiple sources to support the evidence and policy discussion, including reviews of brain pathways in youth, school leaver studies, and cross-national analyses of minimum purchasing age policies. Key references include studies on brain development and alcohol, schoolies experiences, and evaluations of minimum drinking age laws.
Summary of key numerical and statistical references (quick references)
Binge drinking definition used: 5+5+ standard drinks per occasion.
Youth survey on Gold Coast (2010):
74.8extextperthousand74.8extextperthousand played drinking games
64.1extextperthousand64.1extextperthousand consumed >10>10 drinks per night
18.3extextperthousand18.3extextperthousand had sex without protection
13.9extextperthousand13.9extextperthousand had sex with multiple partners
Crashes: lowering age from 2121 to 1818 associated with a 10 ext{%} increase in crashes for 18–2018–20-year-olds.
US history: 2929 states lowered from 2121 to 1818 (1970–1975); federal highway funding consequence led to all states adopting age-21 laws by 19881988.
Underage crash reductions: average reduction around 16 ext{%} across 17 studies.
Support trend: 40.7 ext{%} (2004) to 50.2 ext{%} (2010).
Trickle-down effects observed where harms appeared in younger youths after policy changes.
Canadian and Australian evidence on hospitalisations and traffic injuries shows reductions with higher minimum ages.
Key takeaways for exam preparation
Raising the minimum purchasing age to 2121 is supported by multiple national contexts as a means to reduce youth alcohol harm.
Neurodevelopmental considerations strengthen the argument that 18-year-olds are not fully mature and remain vulnerable to alcohol’s damaging effects.
A comprehensive policy approach, with strong enforcement and public health advocacy, is likely more effective than isolated measures.
Anticipated objections (autonomy, public support, drug substitution, and relevance to targeted road safety) have evidence-based counterarguments that the authors outline.
A strategic advocacy plan emphasizes stakeholder endorsements, ongoing research dissemination, and phased jurisdictional adoption.
READING 9:
Notes on the study: Impact of increasing MLDA from 18 to 20 in Lithuania on all-cause mortality (ITS analysis)
Background
Lithuania increased the minimum legal drinking age (MLDA) from 18 to 20 on January 1, 2018. This change was accompanied by enhanced enforcement requiring retailers to verify age if a customer appears younger than 25 years (Misčikienė et al., 2020).
Other concurrent alcohol control measures enacted to reduce availability and marketing included a ban on marketing and increased enforcement of restrictions as part of a WHO initiative to implement all three WHO ‘best buys’: taxation increases, availability restrictions, and ban of marketing (World Health Organization, 2017; Rehm et al., 2019).
MLDA is defined as the minimum age to purchase and to consume alcohol under law (World Health Organization, 2021).
The policy change raised MLDA to one of the higher thresholds in the WHO European region (only Kazakhstan and Turkmenistan have higher MLDAs at 21). The age restriction applied to both purchasing and consuming alcohol in Lithuania (Misčikienė et al., 2020).
Rationale and expectations: increasing MLDA would be expected to reduce mortality in 18–19-year-olds, with less pronounced or no change in neighboring age groups (15–17 and 20–22) where the MLDA change does not apply.
MLDA aims include reducing drinking and alcohol-attributable problems; most MLDA literature shows reductions in alcohol-related harms when MLDA is increased (US experience 18→21; DeJong et al., 2014; Wagenaar et al., 2002).
Enforcement is crucial for effectiveness (Wagenaar et al., 2005).
International evidence: outside North America, increases in MLDA have been linked to reduced consumption and alcohol-related harm, while lowering MLDA increases risk (Gruenewald et al., 2015; Huckle & Parker, 2014; Jiang et al., 2015).
Objective of the study: estimate the impact of the 2018 MLDA increase on all-cause mortality among 15–22-year-olds in Lithuania, with the hypothesis that the increase would reduce mortality, particularly among 18–19-year-olds.
Design approach: use interrupted time-series analysis (ITS) and compare against control age groups (15–17 and 20–22) to isolate the MLDA effect from other alcohol policies and secular mortality trends. Sensitivity analyses included GDP as a covariate and a taxation policy variable.
Expected outcome: reduction in all-cause mortality among 18–19-year-olds; potential stronger effects in men than women.
Aims and hypotheses
Primary aim: assess whether the 2018 MLDA increase was associated with a decrease in all-cause mortality among 18–19-year-olds in Lithuania.
Secondary aims:
Test whether 15–17-year-olds (controls) and 20–22-year-olds (also affected by other policies) show different patterns.
Evaluate whether inclusion of GDP and alcohol taxation confounds the MLDA effect.
Explore sex-specific effects (men vs women) for the 18–19 age group.
Methods
Data source and period
Monthly all-cause mortality data (monthly number of deaths) for 2001–2019 (n = 228 months) from Statistics Lithuania, provided by the Lithuanian University of Health Sciences.
Mortality data were separated into three age categories: 15–17 (control time series), 18–19 (time series of interest), 20–22 (control time series), and by sex.
Yearly population data for ages 15–22 were obtained and linearly imputed to monthly population data for rate calculation.
GDP per capita (USD) quarterly data were obtained and converted to monthly data via linear imputation to control for economic confounds.
Data on alcohol tax policy: March 2017 exise tax increases (wine and beer up 111%–112%; ethyl alcohol up 23%) used as a potential confound (Misčikienė et al., 2020; Rehm et al., 2021).
Outcome variable
Dependent variable: crude monthly mortality rate per 100,000 individuals for each age group and sex:
Mortality rate formula: ext{MortalityRate}{t} = rac{ ext{Deaths}{t}}{ ext{Population}_{t}} imes 10^5
-Rates are not age-adjusted due to small group sizes.
Primary predictor and covariates
Primary exposure: MLDA policy indicator for January 1, 2018 (Policy2018) coded as 0 before 2018-01-01 and 1 thereafter.
Covariates include seasonality (modeled with a 12-knot cubic spline) and time-series structure to account for autocorrelation.
Economic covariates: GDP per capita (monthly, USD); Alcohol taxation policy (dummy variable for March 2017 tax increase).
Analytical approach
Interrupted time-series analysis (ITS) using a generalized additive mixed model (GAMM) to test the effect of the 2018 MLDA policy, with seasonality controlled by a smoothing spline.
If needed, ARIMA errors were used to achieve stationarity and remove autocorrelation (ARIMA(p,d,q)). Details on model specifications are provided in supplementary materials (Tables S2–S5).
Multiple model specifications:
Model 1: Mortality ~ Intercept + Policy2018 + Seasonality + ARIMA errors (testing MLDA effect alone).
Model 2: Mortality ~ time series differencing with control groups (18–19 minus 15–17; 18–19 minus 20–22; and male minus female within 18–19) to isolate MLDA effect from general mortality trends and co-occurring policies.
Model 3: Add GDP per capita as covariate; evaluate whether the MLDA effect persists.
Model 4: Add taxation policy as covariate; evaluate persistence of MLDA effect.
Period definitions for sensitivity analyses
Primary periods for policy intensity:
Period 1: January 2001 – December 2007 (no major policies).
Period 2: January 2008 – December 2009 (year of sobriety; several policies).
Period 3: January 2010 – February 2017 (no major policies).
Period 4: March 2017 – December 2019 (major alcohol policy implementations; total nine policies).
Sensitivity analyses used alternative period delineations focusing on high-intensity policy windows (Period 3: Jan 2010–Feb 2017; Period 4: March 2017–Dec 2019).
Model evaluation
Assessment of normality and stationarity via QQ plots and the augmented Dickey-Fuller (ADF) test.
Model fit assessed with Adjusted R-squared (R2_adj) and statistical significance of coefficients (P-values).
Bonferroni correction applied for multiple comparisons: with three age groups and multiple model specifications, the adjusted alpha threshold was P < 0.0056.
Software
All analyses conducted in R version 4.0.4 (R Core Team, 2021).
Results
Descriptive findings
Overall, a general decline in all-cause mortality across all three age groups since January 2001.
Injuries accounted for the majority of deaths in 18–19-year-olds (70.01%), with suicide contributing a substantial portion of intentional injury deaths in Lithuania.
ITS results (policy effect alone)
In the most parsimonious model (MLDA policy only):
15–17-year-olds: marginal, non-significant effect (t(226) = −1.76, P = 0.079).
18–19-year-olds: significant effect (t(226) = −3.05, P = 0.0025).
20–22-year-olds: significant effect (t(226) = −3.65, P = 0.00035).
Model details and additional model results are provided in Supplementary Materials (Tables S2–S5).
ITS results with direct control by other time series (differenced time series)
Table 1 (examples): primary effects for policy 2018 across age groups:
18–19-year-olds: Intercept 8.23; Policy2018 −3.07; SE 1.00; t = −3.06; P = 0.0025; Seasonality P = 0.062; AR(1) MA(0).
15–17-year-olds: Intercept 4.41; Policy2018 −1.20; SE 0.68; t = −1.76; P = 0.079; Seasonality significant; AR(5) MA(1).
20–22-year-olds: Intercept 9.46; Policy2018 −3.65; SE 1.00; t = −3.64; P = 0.00033; Seasonality significant; AR(0) MA(2).
Table 2 (differences controlling for other time series): examples:
Total, 18–19 minus 15–17: Intercept 3.82; Policy2018 −1.70; SE 0.85; t = −1.98; P = 0.049; Seasonality not significant; AR(0) MA(1).
Men, 18–19 minus 15–17: Intercept 6.56; Policy2018 −3.62; SE 1.51; t = −2.39; P = 0.018; Seasonality not significant; AR(0) MA(1).
Men, 18–19 minus women, 18–19: Intercept 3.82; Policy2018 −5.73; SE 2.10; t = −2.73; P = 0.0069; Seasonality P < 0.0008; AR(0) MA(1).
Total, 18–19 minus 20–22: Intercept 9.48; Policy2018 −0.73; SE 0.88; t = −0.84; P = 0.40; Seasonality not significant; AR(0) MA(1).
Table 3 (four policy-periods; trends):
Model (total, 18–19): Intercept 8.95; Period 1: 0.013; Period 2: −0.041; Period 3: −0.062; Period 4: −0.080; Period 4 t = −4.80; P < 0.0001 for Seasonality; AR(1) MA(0).
Periods indicate a shift from relatively flat or increasing mortality to a pronounced decline in Period 4 (April 2014–December 2019) with stronger declines in mortality during high-intensity policy windows.
GDP and taxation adjustments
When GDP per capita was included as a covariate, GDP became significant and the MLDA effect disappeared for all three age groups.
When taxation policy (March 2017 tax increases) was added, there was no standalone MLDA effect on mortality for 18–19 nor for the other age groups; however, the 15–17 group showed a marginal effect for taxation (t(224) = −1.90, P = 0.059) and the 20–22 group showed a significant effect (t(224) = −2.24, P = 0.026).
Across ITS models, GDP per capita had a significant negative effect on mortality rates (higher GDP, lower mortality).
Model fit and interpretation
Across GDP- and tax-adjusted models, model fits were relatively modest: approximately 22%–41% of variance explained (Adjusted R²):
15–17-year-olds: R2_adj ≈ 0.22
18–19-year-olds: R2_adj ≈ 0.27
20–22-year-olds: R2_adj ≈ 0.41
Even with confounders, the MLDA effect on 18–19-year-olds was not consistently robust across all models; the effect often diminished when appropriate controls (e.g., 20–22 age group, GDP) were included.
Period-effects and policy clustering
Periods with intensive alcohol policy implementation (Period 2 and Period 4) were associated with changes in mortality trends for 18–19-year-olds: Period 2 shifted from a slight increase to a decline; Period 4 showed a highly significant decrease in all-cause mortality.
Sensitivity analyses showed similar patterning; the broader policy implementation periods amplified the observed downward mortality trend, especially among men.
Overall conclusion from results
There was a general decline in all-cause mortality among young Lithuanians and an association with alcohol policy implementation in the period; however, the exact causal contribution of the MLDA change is unclear.
When controlling for other alcohol policy measures and macroeconomic factors, the isolated MLDA effect was not definitively demonstrated as causal.
The direction of effects is consistent with prior MLDA literature, but a definitive causal link between MLDA increase and reductions in all-cause mortality could not be established within the short post-policy window (about one year) and given confounding policy clusters.
Discussion
Interpretation of findings
A general downward trend in mortality among 18–19-year-olds followed the broader rollout of alcohol control policies; the naked MLDA effect, when tested alone, appeared significant, but failed to remain significant after including control series and economic confounds.
The persistent but non-definitive signal aligns with decades of MLDA research showing inverse relationships between MLDA and alcohol-related harms, though attribution to MLDA alone remains difficult.
The results are compatible with indirect MLDA effects, such as changes in drinking patterns and risk-taking behaviors, but the near-simultaneous implementation of multiple policies complicates causal attribution to MLDA alone.
Limitations acknowledged by authors
Lithuania is a small country; monthly death counts per subgroup are low, leading to unstable mortality rate estimates and a relatively poor model fit.
Population registration changes reduced recorded population over the study period, potentially inflating fluctuations in mortality rates when deaths are few.
Data limitations prevented separating male and female deaths for all models; small numbers hindered sex-specific analyses in some strata.
Alcohol policy implementation occurred in clusters; disentangling the isolated impact of MLDA from marketing ban, availability restrictions, and taxation changes is difficult, particularly since there is strong temporal correlation among policies (Pearson r ≈ 0.81 between policy timing and policy intensity).
Possible spillover effects: policy announcements may influence behavior in adjacent age groups before official policy enactment; cross-age interdependencies (e.g., group drinking dynamics) could dilute age-specific effects.
Multiple model runs increased the risk of Type I error; a Bonferroni-corrected threshold was applied, and the primary 18–19-year-old MLDA effect remained significant under that correction (P = 0.0025 < 0.0056).
Context with prior research
The study’s direction and qualitative findings are consistent with prior MLDA literature demonstrating that stronger alcohol-control policies reduce alcohol-related harms and all-cause mortality, though the isolated impact of MLDA remains debated in cross-national contexts.
The authors emphasize that, despite lack of a definitive causal link to MLDA in this study, the broader policy environment during 2014–2019 likely contributed to mortality declines via multiple channels (e.g., reductions in drinking, binge drinking, and risky behaviors).
Implications for policy and future research
The findings support continued adoption of comprehensive alcohol-control policies in Lithuania and similar settings, while recognizing the challenges in isolating the causal impact of any single policy like MLDA.
Future research with longer follow-up, larger populations, and more granular data (e.g., cause-specific mortality and injuries, age-by-sex stratifications) could help clarify the isolated effect of MLDA.
Investigations into interactions among policies (taxation, advertising bans, availability restrictions) and their combined effects on mortality are warranted.
Limitations (detailed)
Small numbers: mortality counts in 18–19-year-olds are low, leading to unstable rates and difficulties in separating female data.
Population register changes reduced reliability of population denominators across the study window.
Policy clustering: multiple alcohol-control measures implemented contemporaneously; attributing effects to a single policy (MLDA) is difficult due to collinearity and overlapping impact.
Moderate model fit: R²_adj values indicate substantial unexplained variance, reflecting both data limitations and the complexity of policy effects on mortality.
Spillover and inter-age interactions: behavior and drinking contexts can shift across age groups, limiting clean causal inference for a single age band.
Conclusions
There was a general decline in all-cause mortality among Lithuanians aged 15–22 from 2001 to 2019, with mortality reductions temporally associated with broader alcohol-control policies implemented around 2017–2019.
The study did not definitively establish a causal impact of the MLDA increase (from 18 to 20) on all-cause mortality in 18–19-year-olds when controlling for other age groups and macroeconomic factors.
Some analyses suggested an MLDA-associated reduction in mortality for 18–19 and 20–22-year-olds, but these effects diminished once GDP and taxation policies were included, indicating that economic context and other policy changes likely contributed to observed declines.
Overall, while MLDA may play a role in reducing alcohol-related harms, the observed mortality trends in Lithuania during 2001–2019 cannot be attributed solely to the 2018 MLDA policy; the broader slate of alcohol-control policies is more plausibly linked to mortality decreases, with the MLDA contribution remaining uncertain within the study’s design and window.
Supplementary materials and data access
Supplementary material is available online with Alcohol and Alcoholism.
Data availability: R code used for analysis available on request from the corresponding author; data may be obtained through Lithuanian governmental institutions (Lithuanian Institute of Hygiene, Statistics Lithuania).
Acknowledgments, funding, and conflicts of interest
Funding: Research supported by the National Institute on Alcohol Abuse and Alcoholism (NIAAA), NIH (Award Number 1R01AA028224-01).
Additional funding acknowledgments: Jürgen Rehm; JR acknowledges funding from the Canadian Institutes of Health Research, Institute of Neurosciences, Mental Health and Addiction (CRISM Ontario Node grant no. SMN-13950).
Conflicts of interest: None declared. Authors state views do not necessarily represent positions of WHO or other organizations.
References (selected, as cited in the transcript)
Babor, Caetano, Casswell, et al. Alcohol: No Ordinary Commodity. 2nd edn. Oxford University Press, 2010.
Beard, Marsden, Brown, et al. Understanding and using time series analyses in addiction research. Addiction, 2019.
Carpenter, Dobkin. The minimum legal drinking age and public health. J Econ Perspect, 2011; and The minimum legal drinking age and crime. Rev Econ Stat, 2015.
DeJong, Blanchette. Case closed: MLDA 21 impact in USA. J Stud Alcohol Drugs, 2014.
Gruenewald, Treno, Ponicki, et al. Impacts of MLDA changes outside North America. Addiction, 2015.
Huckle, Parker. Long-term impact on alcohol-involved crashes of lowering MLDA in NZ. Am J Public Health, 2014.
Jiang, Livingston, Manton. MLDA effects on traffic fatalities in Australia. Inj Prev, 2015.
Misčikienė, Midttun, Galkus, et al. Review of Lithuanian alcohol control legislation (1990–2020). Int J Environ Res Public Health, 2020.
Neufeld, Bobrova, Davletov, et al. Alcohol control policies in former Soviet countries: a narrative review. Drug Alcohol Rev, 2021.
Rehm, Gmel, Gmel, et al. The relationship between dimensions of alcohol use and burden of disease. Addiction, 2017.
Rehm, Stelemėkas, Badaras, et al. Alcohol control policy measures and mortality in Lithuania: ITS analysis. Addiction, 2021.
Wagenaar, Toomey. Effects of minimum drinking age laws: literature review 1960–2000. J Stud Alcohol Suppl, 2002.
Wagenaar, Toomey, Erickson. Compliance with the minimum drinking age: enforcement and training interventions. Alcohol Clin Exp Res, 2005.
World Health Organization. Best Buys and Other Interventions for NCD Prevention. 2017.
World Health Organization. Global status report on alcohol and health 2018; World Health Organization. Suicide In The World: Global Health Estimates, 2019; WHO. Minimum legal drinking age, 2021.
Numerous peer-reviewed studies have investigated the impact of minimum legal drinking ages (MLDAs) on alcohol-attributable morbidity and mortality, showing varied outcomes depending on the direction of the policy change and the specific context.
Effects of Raising the Minimum Legal Drinking Age (MLDA)
Raising the MLDA, typically from 1821182112 \text{%}1716 \text{%}19842125,000198872.9 \text{%} of studies on secondary societal harm without a bridging variable found significant reductions in harms like traffic outcomes after MLDA increases (Reading 6, Path 3).
Reduced Broader Alcohol-Attributable Morbidity and Mortality:
Benefits extend beyond road safety to encompass reductions in alcohol dependence, alcohol-related violence, suicide, and risky sexual behaviors among youth (Reading 7).
Canadian data showed that a higher MLPA reduced youth hospitalization rates for alcohol-use disorders, alcohol poisoning, suicidal behavior, and traffic crash injuries (Reading 8).
The scoping review (Reading 6, Path 3 & Path 4) also found several studies reporting reductions in non-traffic harms such as teen suicide, sexually transmitted diseases, and birth outcomes.
Reductions in Youth Drinking Patterns:
Raising the MLDA has been linked to decreased alcohol use among 18 \text{-} 2115 \text{-} 18211821+27 \text{-} 632125182021181970197518 \text{-} 2010 \text{%}2118196915 \text{-} 17201820181999$$ did not lead to an immediate, noticeable increase in alcohol-related or total vehicular accidents among teens in the short run or in cumulative terms for affected cohorts (Reading 2). This notable exception highlights the importance of specific contextual factors, enforcement, and public discussion in shaping policy outcomes.
peer reviewed artciles
- effects of raising/lowering on alcohol consumption patterns among young people - e.g., will they still drink when it stops being legal for them
Numerous peer-reviewed studies have investigated the impact of minimum legal drinking ages (MLDAs) on alcohol consumption patterns among young people. Generally, raising the MLDA tends to reduce alcohol use, while lowering it can lead to increased consumption, though behavioral adaptations and contextual factors play a significant role in how young people access and consume alcohol regardless of legality.
Effects of Raising the Minimum Legal Drinking Age (MLDA)
Raising the MLDA, typically from 1818 to 2121, has been associated with reductions in various aspects of alcohol consumption among young people, but also with shifts in where and how they obtain alcohol:
Reduced Overall Alcohol Consumption and Initiation:
Many studies show a negative (protective) impact on primary drinking patterns, including measures like past-month use, number of drinking occasions, and binge drinking in the short term, and some long-term reductions (Reading 6).
Upward shifts in the MLDA reduce drinking among 18-2118-21-year-olds and also reduce initiation among 15-1815-18-year-olds, suggesting that the policy effects transmit to younger cohorts (Reading 7).
Cross-country comparisons, such as U.S. data (with an MLDA of 2121) compared to many European countries (with lower MLDAs), show lower past-30-day drinking rates and a lower proportion of youth intoxicated before age 1313 in the U.S. (Reading 7). This implies a broader cultural shift towards less drinking when the MLDA is higher.
Shift in Drinking Patterns and Sources (Addressing Continued Drinking):
While raising the MLDA aims to reduce consumption, evidence suggests that young people may adapt their behaviors rather than ceasing drinking entirely when it becomes illegal for them.
Studies have observed changes in purchasing behavior, with teens less likely to purchase from legal outlets and more likely to obtain alcohol via parties or older friends and siblings (Reading 6, Reading 7).
"Substitution effects" have been noted, where some youth may alter where and how they obtain alcohol, such as shifting to home or other private settings for consumption (Reading 6). For instance, Dutch data indicated that private property drinking by 16-1716-17-year-olds increased after the MLDA was raised (Reading 6).
Despite concerns that stricter laws could push drinking "underground" and promote unsafe patterns (Reading 7), studies generally show that a higher MLDA still leads to overall reductions in consumption and related harms. There is no consistent evidence that raising the MLDA leads to a compensatory rise in illicit drug use (Reading 8).
Effects of Lowering the Minimum Legal Drinking Age (MLDA)
Conversely, lowering the MLDA is generally associated with increased alcohol consumption and a "trickle-down" effect to younger age groups:
Increased Consumption and "Trickle-Down" Effects:
Lowering the legal drinking age has been linked to higher youth harm and an observed "trickle-down effect," where harms and access to alcohol can increase among even younger adolescents (e.g., 15-1715-17-year-olds) due to easier access through older peers and siblings who can legally purchase alcohol (Reading 7, Reading 8).
New Zealand Exception (MLDA 2020 to 1818):
A significant exception is seen in New Zealand, where lowering the MLDA from 2020 to 1818 in 19991999 did not lead to an immediate or long-term increase in alcohol-related or total vehicular accidents, nor did it result in a large long-run shift in youth drinking patterns across affected cohorts (Reading 2).
While short-run increases in risky drinking were observed among teenagers immediately after gaining legal access, long-run drinking patterns converged across cohorts. This unique outcome highlights the importance of specific contextual factors, enforcement, and public discussion in mediating policy impacts on consumption (Reading 2).
In summary, while raising the MLDA effectively reduces overall youth alcohol consumption and initiation, young people may still engage in drinking by adapting their sourcing and location to skirt the law. Conversely, lowering the MLDA often leads to increased