AIを用いて大麻政策に対する世論を分析(Study: Americans support cannabis rescheduling)

2026-04-29 ジョンズ・ホプキンス大学(JHU)

ジョンズ・ホプキンス大学の研究は、米国における大麻の規制区分見直し(再スケジューリング)に対する国民の支持を調査した。調査結果から、多くの米国人が医療用途などを踏まえた規制緩和に賛成していることが明らかとなり、政策変更への社会的受容が高まっていることが示された。一方で、安全性や依存性に対する懸念も依然として存在し、意見は一様ではない。研究は、世論が薬物政策に与える影響の重要性を示し、今後の制度設計や公衆衛生議論に示唆を与える。

AIを用いて大麻政策に対する世論を分析(Study: Americans support cannabis rescheduling)

<関連情報>

米国における大麻の規制変更案に対するRegulations.govを通じた一般からの意見を分析する Characterizing public comments via Regulations.gov in response to proposed cannabis rescheduling in the United States

Vijay M. Tiyyala, Cerina Dubois, Clarissa Madar, Ryan Vandrey, Johannes Thrul, Mark Dredze, John W. Ayers

Addition  Published: 24 April 2026

DOI:https://doi.org/10.1111/add.70410

Abstract

Aims

The United States Drug Enforcement Administration’s (DEA) proposed rescheduling of cannabis from Schedule I to Schedule III under the Controlled Substances Act marks a significant shift in federal policy. Understanding public sentiment toward this policy is critical for guiding the current cannabis rescheduling effort as well as future reforms. The objective of this study is to characterize public comments submitted to Regulations.gov regarding the DEA’s cannabis rescheduling proposal and identify underlying justifications for support or opposition.

Design

A mixed-methods analysis was conducted.

Setting

Online public comments submitted to Regulations.gov regarding the DEA’s cannabis rescheduling proposal.

Participants

42 913 public comments submitted between 21 May and 22 July 2024.

Measurements

Comments were analyzed for sentiment towards the proposed rescheduling (support, oppose or insufficient rescheduling) and thematic justifications using manual and automated natural language processing techniques. A two-stage annotation approach was employed: manual coding of 200 randomly sampled comments by multiple independent evaluators, followed by automated classification of all 42 913 comments using open source Large Language Model (LLM) validated against the manual annotations.

Findings

Using LLM-based classification validated against human annotations [88% agreement, F1 (harmonic mean of precision and recall) = 0.86], we found that among 42 913 comments, 28.85% [95% confidence interval (CI) = 28.44%–29.24%] supported rescheduling, 6.74% (95% CI = 6.50%–6.99%) opposed and 63.50% (95% CI = 63.06%–63.99%) deemed the proposal insufficient, favoring further rescheduling or complete de-scheduling of cannabis. Among the 200 manually annotated comments, therapeutic benefits (56.7%, 95% CI = 46.7%–66.7%) and economic impacts (27.8%, 95% CI = 18.9%–37.8%) were the most common justifications among supporters. Public health risks (100.0%, 95% CI = 100.0%–100.0%), addictiveness concerns (71.4%, 95% CI = 42.9%–100.0%) and concerns about underage use (57.1%, 95% CI = 14.3%–85.7%) were predominant in opposing comments. Insufficient rescheduling comments cited therapeutic benefits (37.8%, 95% CI = 28.5%–48.0%), economic impacts (28.6%, 95% CI = 19.4%–37.8%) and criminal justice reform (26.5%, 95% CI = 18.4%–35.7%) as primary justifications.

Conclusions

Public sentiment on Regulations.gov supports the United States Drug Enforcement Administration’s proposal for cannabis rescheduling, though the majority views the proposed Schedule III classification as inadequate and supports further rescheduling or complete de-scheduling of cannabis.

1504数理・情報
ad
ad
Follow
ad
タイトルとURLをコピーしました