서브메뉴
검색
A Mixed-Methods Procedure for Creating and Testing Idiographic and Pooled Prediction Models of In-the-Moment Cannabis Use
A Mixed-Methods Procedure for Creating and Testing Idiographic and Pooled Prediction Models of In-the-Moment Cannabis Use
상세정보
- 자료유형
- 학위논문 서양
- 최종처리일시
- 20260202104816
- ISBN
- 9798297601253
- DDC
- 150
- 서명/저자
- A Mixed-Methods Procedure for Creating and Testing Idiographic and Pooled Prediction Models of In-the-Moment Cannabis Use
- 발행사항
- [Sl] : University of California, Berkeley, 2025
- 발행사항
- Ann Arbor : ProQuest Dissertations & Theses, 2025
- 형태사항
- 86 p
- 주기사항
- Source: Dissertations Abstracts International, Volume: 87-04, Section: B.
- 주기사항
- Includes supplementary digital materials.
- 주기사항
- Advisor: Fisher, Aaron.
- 학위논문주기
- Thesis (Ph.D.)--University of California, Berkeley, 2025.
- 초록/해제
- 요약The rising prevalence of cannabis use (and associated negative consequences) necessitates a deeper understanding of the real-time, person-specific mechanisms driving in-the-moment substance use. Traditional nomothetic research, while valuable for identifying group-level risk factors, often fails to capture the idiographic dynamics that are most relevant for personalized intervention. This dissertation outlines and validates a comprehensive, four-study, mixed-methods framework for developing and evaluating high-accuracy, personalized prediction models of in-the-moment cannabis use.First, a systematic methodological review of 54 ecological momentary assessment (EMA) studies revealed a field characterized by profound methodological heterogeneity and a pervasive lack of reporting transparency, establishing the need for a more rigorous, foundational approach to study design. Second, these findings informed a community-partnered participatory design study (N=496) that synthesized academic precedent with the lived experience of cannabis users to develop a feasible, relevant, and ecologically valid EMA protocol. Third, data from an intensive longitudinal study of 113 adults who regularly use cannabis were used to train idiographic prediction models using an elastic net machine learning algorithm. These personalized models demonstrated exceptional predictive accuracy (mean AUC = .86) and were found to be vastly superior to traditional pooled nomothetic models and participant-generated heuristic models. A key finding emerged: participants identified static, contextual factors as their primary triggers, whereas the algorithm achieved superior performance by prioritizing dynamic, internal affective states.Finally, a multi-faceted Bayesian comparison framework was developed to empirically quantify the trustworthiness of each idiographic model. While the elastic net models were predictively superior to their Bayesian counterparts for a large majority of the sample (89.4%), structural analyses revealed a consistent pattern of thematic convergence despite low variable-level agreement. This work makes three primary contributions: it provides a replicable, end-to-end pipeline for personalized science; it offers a nuanced, data-driven model of cannabis use that highlights the limits of conscious self-reflection; and it establishes a novel, empirical rubric for assessing model uncertainty. This dissertation provides a rigorous methodological roadmap for the development of personalized science and lays the groundwork for translating these predictive models into targeted, just-in-time adaptive interventions.
- 일반주제명
- Psychology
- 일반주제명
- Statistics
- 일반주제명
- Biology
- 키워드
- Cannabis use
- 키워드
- Idiographic
- 기타저자
- University of California, Berkeley Psychology
- 기본자료저록
- Dissertations Abstracts International. 87-04B.
- 전자적 위치 및 접속
- 로그인 후 원문을 볼 수 있습니다.
MARC
008260126s2025 us c eng d■001000017358970
■00520260202104816
■006m o d
■007cr#unu||||||||
■020 ▼a9798297601253
■035 ▼a(MiAaPQ)AAI32168564
■040 ▼aMiAaPQ▼cMiAaPQ
■0820 ▼a150
■1001 ▼aSoyster, Peter D.
■24512▼aA Mixed-Methods Procedure for Creating and Testing Idiographic and Pooled Prediction Models of In-the-Moment Cannabis Use
■260 ▼a[Sl]▼bUniversity of California, Berkeley▼c2025
■260 1▼aAnn Arbor▼bProQuest Dissertations & Theses▼c2025
■300 ▼a86 p
■500 ▼aSource: Dissertations Abstracts International, Volume: 87-04, Section: B.
■500 ▼aIncludes supplementary digital materials.
■500 ▼aAdvisor: Fisher, Aaron.
■5021 ▼aThesis (Ph.D.)--University of California, Berkeley, 2025.
■520 ▼aThe rising prevalence of cannabis use (and associated negative consequences) necessitates a deeper understanding of the real-time, person-specific mechanisms driving in-the-moment substance use. Traditional nomothetic research, while valuable for identifying group-level risk factors, often fails to capture the idiographic dynamics that are most relevant for personalized intervention. This dissertation outlines and validates a comprehensive, four-study, mixed-methods framework for developing and evaluating high-accuracy, personalized prediction models of in-the-moment cannabis use.First, a systematic methodological review of 54 ecological momentary assessment (EMA) studies revealed a field characterized by profound methodological heterogeneity and a pervasive lack of reporting transparency, establishing the need for a more rigorous, foundational approach to study design. Second, these findings informed a community-partnered participatory design study (N=496) that synthesized academic precedent with the lived experience of cannabis users to develop a feasible, relevant, and ecologically valid EMA protocol. Third, data from an intensive longitudinal study of 113 adults who regularly use cannabis were used to train idiographic prediction models using an elastic net machine learning algorithm. These personalized models demonstrated exceptional predictive accuracy (mean AUC = .86) and were found to be vastly superior to traditional pooled nomothetic models and participant-generated heuristic models. A key finding emerged: participants identified static, contextual factors as their primary triggers, whereas the algorithm achieved superior performance by prioritizing dynamic, internal affective states.Finally, a multi-faceted Bayesian comparison framework was developed to empirically quantify the trustworthiness of each idiographic model. While the elastic net models were predictively superior to their Bayesian counterparts for a large majority of the sample (89.4%), structural analyses revealed a consistent pattern of thematic convergence despite low variable-level agreement. This work makes three primary contributions: it provides a replicable, end-to-end pipeline for personalized science; it offers a nuanced, data-driven model of cannabis use that highlights the limits of conscious self-reflection; and it establishes a novel, empirical rubric for assessing model uncertainty. This dissertation provides a rigorous methodological roadmap for the development of personalized science and lays the groundwork for translating these predictive models into targeted, just-in-time adaptive interventions.
■590 ▼aSchool code: 0028.
■650 4▼aPsychology
■650 4▼aStatistics
■650 4▼aBiology
■653 ▼aCannabis use
■653 ▼aCommunity partners
■653 ▼aEcological momentary assessment
■653 ▼aIdiographic
■653 ▼aPrediction models
■690 ▼a0621
■690 ▼a0306
■690 ▼a0463
■71020▼aUniversity of California, Berkeley▼bPsychology.
■7730 ▼tDissertations Abstracts International▼g87-04B.
■790 ▼a0028
■791 ▼aPh.D.
■792 ▼a2025
■793 ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T17358970▼nKERIS▼z이 자료의 원문은 한국교육학술정보원에서 제공합니다.
