{"id":397,"date":"2026-07-06T14:19:44","date_gmt":"2026-07-06T14:19:44","guid":{"rendered":"https:\/\/openpub.libraries.rutgers.edu\/dsarm12\/?post_type=chapter&p=397"},"modified":"2026-07-06T21:32:22","modified_gmt":"2026-07-06T21:32:22","slug":"social-and-environmental-pathways-data-quality-and-data-curation","status":"publish","type":"chapter","link":"https:\/\/openpub.libraries.rutgers.edu\/dsarm12\/chapter\/social-and-environmental-pathways-data-quality-and-data-curation\/","title":{"raw":"Social and Environmental Pathways, Data Quality, and Data Curation","rendered":"Social and Environmental Pathways, Data Quality, and Data Curation"},"content":{"raw":"

Reading Objectives<\/h2>\r\nBy the end of this chapter, you should be able to:\r\n
    \r\n \t
  1. Describe social and environmental conditions<\/strong> that may be associated with adolescent substance use, including attitudes, family relationships, peers, schools, neighborhoods, and availability of substances.<\/li>\r\n \t
  2. Identify examples of instruments<\/strong> that measure attitudinal and environmental constructs, and recognize why multiple reporters and assessment waves are used.<\/li>\r\n \t
  3. Explain why data quality and transparent cleaning decisions matter<\/strong> before analyzing social and environmental data.<\/li>\r\n \t
  4. Distinguish conceptually among MCAR, MAR, and MNAR missingness mechanisms<\/strong>, and understand why these labels are assumptions rather than provable facts.<\/li>\r\n \t
  5. Describe the source\u2010to\u2010analytic data workflow<\/strong>, including identifiers, events, versions, and documentation.<\/li>\r\n \t
  6. Recognize ethical implications of cleaning and interpreting social and environmental data<\/strong>, including risks of exclusion, over\u2011cleaning, stigma, privacy loss, and unsupported causal claims.<\/li>\r\n<\/ol>\r\n
    \r\n

    Key Terms<\/h2>\r\n

    Attitudinal factors; data quality; environmental factors; longitudinal data; missingness (MCAR, MAR, MNAR); quality-control flags; relational data; skip logic.<\/p>\r\n\r\n<\/div>\r\n

    3.1\u00a0Social and Environmental Pathways to Substance Use<\/h2>\r\nIn the 1970s and early 1980s, psychologist Bruce Alexander and colleagues used a set of animal experiments often called the Rat Park studies to challenge a simplified view of addiction as only drug-driven. Across these studies, rats housed in larger, socially enriched environments often consumed less morphine solution than rats housed in more restricted laboratory housing (Alexander et al., 1981).<\/strong>\r\n\r\nhttps:\/\/www.youtube.com\/watch?v=d-0KfwFCMRM\r\n\r\nRat Park does not provide a direct explanation of human substance use, and it should not be treated as a single definitive experiment. However, it offers a useful starting question: how might social connection, stress, opportunity, and environment shape substance-related behavior?\r\n\r\nIn human development, substance use is likewise associated with multiple interacting conditions. Young people\u2019s attitudes, family relationships, peer networks, school experiences, neighborhood conditions, and access to substances can shape opportunities, expectations, stressors, and supports. These conditions do not determine any individual\u2019s outcome, but they help researchers investigate why pathways into and away from substance use differ across people and over time.\r\n\r\nResearchers speak of pathways<\/em> because there are many possible combinations of influences rather than one deterministic sequence. A young person\u2019s beliefs about the effects of a substance\u2014often called attitudes or expectancies<\/strong>\u2014may be shaped by what they hear from peers, family members, media, or their neighborhood. These beliefs interact with family relationships, school engagement, neighborhood safety, and access to substances. A pathway is a pattern to investigate, not a prediction about any individual.\r\n

    Attitudes, Expectations, and Perceived Harm<\/h3>\r\nYoung people often form beliefs about substances long before they have direct experience with them. Expectancies <\/strong>describe beliefs about the effects a substance might have, such as feeling more relaxed or more confident. Motives<\/strong> capture reasons for considering or using a substance, such as coping, curiosity, or social connection. Intentions<\/strong> refer to plans or expectations to use a substance in the future, and perceived harm<\/strong> reflects beliefs about the risks of use. These constructs are latent: they cannot be directly observed but must be inferred from survey questions. Positive expectancies, low perceived harm, curiosity, and intentions may be associated with experimentation, whereas perceiving greater risk may be associated with lower willingness to initiate use. Attitudes themselves may reflect broader social norms and can change after someone gains experience with a substance.\r\n

    Family Environments<\/h3>\r\nFamily conditions can provide support, structure, or stress. Parental monitoring<\/strong> refers to how much caregivers know about a young person\u2019s activities and friends. Monitoring is meaningful only in context: it can reflect open communication and shared routines, or it may result from enforcement and surveillance. Family cohesion<\/strong> describes support and emotional closeness, whereas family conflict<\/strong> refers to recurring tension or hostility. These constructs matter because support can buffer stress and shape opportunities, while conflict or lax monitoring may coincide with riskier contexts. These conditions may be associated with different substance-use pathways, although they do not determine any individual\u2019s outcome (Buu et al., 2009). In ABCD, measures such as the Parental Monitoring scale and the Family Environment Scale provide structured evidence of these aspects but do not capture every nuance of family life.\r\n

    Peers, Social Norms, and Resistance to Influence<\/h3>\r\nPeer contexts may be associated with substance use in multiple ways. Friendship networks, shared activities, and social norms can increase exposure to substances or create expectations about use (Curran et al., 1997; Dielman et al., 1990). At the same time, young people often select friends who share their interests or experiences, and both peer relationships and substance use may be shaped by family, school, or neighborhood conditions. Resistance to peer influence<\/strong> is a person\u2019s ability to maintain an independent decision when feeling social pressure (Steinberg & Monahan, 2007). ABCD measures such as the Peer Behavior Profile and Resistance to Peer Influence capture aspects of peer context and the capacity to resist pressure, helping researchers examine whether peer environments are associated with substance use while recognizing their complexity.\r\n

    School Engagement and Context<\/h3>\r\nSchools are not just academic settings; they are major social environments. School engagement<\/strong> includes participation in activities, connections with teachers or classmates, interest in learning, and attendance. School context<\/strong> encompasses climate, relationships, resources, and policies. Engagement can provide structure and support, but it is shaped by factors beyond individual control\u2014such as transportation, exclusionary discipline, or resource availability. Measures like the School Risk and Protective Factors scale assess how engaged students feel and whether school contexts provide support or pose risks.\r\n

    Neighborhood Conditions and Substance Availability<\/h3>\r\nNeighborhoods shape daily experiences through safety, social cohesion, transportation, housing, recreation, stressors, and local policies. A young person\u2019s perception of neighborhood safety<\/strong> may affect where they spend time and how they experience stress. Substance availability<\/strong> refers to how easily substances may be obtained at home, in peer groups, in school, or in the community. Availability alone does not determine behavior; its effects depend on rules, norms, perceived harm, and individual experiences. Measures like the Neighborhood Safety\/Crime Survey and Community Risk and Protective Factors describe perceived safety, community supports, and access to substances, but they should not be treated as labels for communities or residents. A neighborhood\u2011safety score reflects a particular perception at a particular time, not a complete description of a place.\r\n

    Pathways Are Dynamic<\/h3>\r\nThese social and environmental factors operate together and change across development. Substance use can also alter later relationships, attitudes, and engagement. For example, experiences with a substance may influence future beliefs, peer networks, family conflict, or school involvement. Researchers therefore need longitudinal data<\/strong>\u2014repeated observations of the same participants over time\u2014to examine how attitudes, contexts, and behaviors co\u2011evolve. Recognizing that pathways are non\u2011deterministic and context\u2011dependent helps prevent stigma and avoids simplistic explanations for substance use.\r\n\r\n \r\n\r\n[caption id=\"attachment_401\" align=\"aligncenter\" width=\"1024\"]\"Conceptual Figure 3.1. Social and environmental pathways to youth substance use[\/caption]\r\n

    3.2\u00a0Measurement of Attitudes and Environments<\/h2>\r\n

    As Chapter 2 introduced, many important research constructs and concepts are not directly observed as a single raw value. Perceived harm, family cohesion, peer norms, school engagement, and neighborhood conditions are measured through indicators such as survey items, caregiver reports, administrative records, or composite scores.<\/p>\r\n

    The same measurement questions apply here: What was the instrument designed to capture? Who provided the information? When was it collected? How were items scored? Which responses were skipped, missing, or not applicable? Before using an attitude or environmental variable in analysis, researchers must consult the relevant documentation and interpret the value in light of how it was produced.<\/p>\r\n\r\n

    Indicators, Scales, and Summary Scores<\/h3>\r\nMost attitudinal and environmental measures in the ABCD Study are implemented as scales<\/strong> consisting of multiple items. Each item provides a piece of evidence about the underlying construct. Analysts often combine items into a summary score<\/strong> following documented rules\u2014sometimes reversing the direction of particular items or requiring that a minimum number of items be present. For example, a Parental Monitoring scale might ask how often a caregiver knows where the young person is, who their friends are, or what they do after school. These items are then combined into a score that represents perceived monitoring.\r\n

    Multiple Reporters and Informant Differences<\/h3>\r\nMany social experiences can be described from more than one perspective. Youth may be best positioned to report on private experiences, peer contexts, or perceived harm, while caregivers may provide additional information about household routines or observed behavior. Differences between youth and caregiver reports are not automatically errors or attempts to deceive. They may reflect different opportunities to observe an experience, different definitions of behaviors, privacy considerations, or changing circumstances. Consequently, analysts should avoid averaging or combining reports without a clear conceptual justification and documented scoring approach.\r\n

    Examples of Measures<\/h3>\r\nTable 3.1 provides examples of how latent constructs are operationalized in the ABCD Study (Lisdahl et al. 2018). It summarizes selected instruments, reporters, and broad assessment timing. This is not a complete inventory; rather, it highlights the variety of measures and the importance of consulting documentation before combining or comparing variables.\r\n\r\n\r\n\r\n\r\n\r\n\r\n\r\n\r\n\r\n\r\n\r\n\r\n
    Latent construct<\/strong><\/td>\r\nExample measure<\/strong><\/td>\r\nReporter<\/strong><\/td>\r\nIllustrative timing<\/strong><\/td>\r\nWhat it captures<\/strong><\/td>\r\n<\/tr>\r\n<\/thead>\r\n
    Substance expectancies<\/td>\r\nAlcohol, cannabis, nicotine, or vaping expectancy scales<\/td>\r\nYouth<\/td>\r\nBaseline and selected follow\u2011ups<\/td>\r\nBeliefs about effects (e.g., relaxation, social confidence, harm)<\/td>\r\n<\/tr>\r\n
    Motives for use<\/td>\r\nAlcohol, cannabis, nicotine, or vaping motives scales<\/td>\r\nYouth<\/td>\r\nPrimarily later follow\u2011ups<\/td>\r\nReasons for considering or using a substance (e.g., coping, social enhancement)<\/td>\r\n<\/tr>\r\n
    Intentions and perceived harm<\/td>\r\nIntention\u2011to\u2011use and perceived\u2011harm items<\/td>\r\nYouth<\/td>\r\nBaseline and early follow\u2011ups<\/td>\r\nWillingness or plans to use substances and beliefs about risks<\/td>\r\n<\/tr>\r\n
    Parental monitoring<\/td>\r\nParental Monitoring scale<\/td>\r\nYouth (and sometimes caregivers)<\/td>\r\nRepeated annually<\/td>\r\nCaregiver knowledge about activities, whereabouts, and social contexts<\/td>\r\n<\/tr>\r\n
    Family environment<\/td>\r\nFamily Environment Scale<\/td>\r\nYouth<\/td>\r\nRepeated annually<\/td>\r\nCohesion, conflict, and aspects of household emotional climate<\/td>\r\n<\/tr>\r\n
    Peer context<\/td>\r\nPeer Behavior Profile; Protective Peer Network; Resistance to Peer Influence<\/td>\r\nYouth<\/td>\r\nSelected follow\u2011ups<\/td>\r\nPeer behavior, protective networks, and perceived ability to resist pressure<\/td>\r\n<\/tr>\r\n
    School context<\/td>\r\nSchool Risk and Protective Factors scale<\/td>\r\nYouth<\/td>\r\nRepeated annually<\/td>\r\nSchool environment, involvement, connection, and disengagement<\/td>\r\n<\/tr>\r\n
    Neighborhood and community context<\/td>\r\nNeighborhood Safety\/Crime Survey; Community Risk and Protective Factors<\/td>\r\nYouth<\/td>\r\nBaseline and selected follow\u2011ups<\/td>\r\nPerceived safety, community supports, and access to substances<\/td>\r\n<\/tr>\r\n<\/tbody>\r\n<\/table>\r\n

    Documentation and Interpretation<\/h3>\r\nInterpreting these measures requires consulting the data dictionary<\/strong>, codebook<\/strong>, and release notes<\/strong>. These documents identify the item wording, response options, scoring rules, reporter, timeframe, skip logic, and versions. For example, similar constructs may be measured differently across substances or developmental stages; an alcohol expectancy scale and a cannabis expectancy scale are not automatically interchangeable. Analysts must understand which items belong to a scale, whether higher values indicate more or less of the construct, and whether any special codes represent refusal or skip responses. Only then can summary scores be responsibly used or compared across participants.\r\n

    3.3\u00a0Data Quality as Concept and Practice<\/h2>\r\nAttitudinal and environmental measures do not arrive in a dataset as simple, self\u2011explanatory facts. Before linking family, peer, school, neighborhood, and attitudinal measures to substance\u2011use outcomes, researchers must ask whether the available data are suitable for the question they want to answer. Data quality<\/em> is not about perfection; it is about whether data are accurate<\/strong>, complete<\/strong>, consistent<\/strong>, timely<\/strong>, and relevant<\/strong> for a particular analysis. Large longitudinal studies almost always include incomplete records, unusual values, different reporters, technical problems, and changes across study visits. The goal is to understand what the data represent, identify limitations, and make transparent decisions about how to use them.\r\n

    Inspection versus Cleaning<\/h3>\r\nData inspection<\/strong> means examining a dataset to understand its structure, values, patterns, and potential concerns. Analysts read the data dictionary, check ranges and variable types, count missing values, review unusual categories, and create simple summaries. Data cleaning<\/strong> means making documented changes based on what inspection reveals: recoding a special missing\u2011value code, converting a variable to the correct type, creating a quality\u2011control flag, or excluding records according to a justified rule. Inspection should come first. An unusual value is not automatically an error, and a blank cell is not automatically missing data. Rather than overwriting raw values, analysts should create new cleaned variables so that the original information is preserved.\r\n

    Missing, Skipped, and Special\u2011Coded Values<\/h3>\r\nValues may be absent for many reasons: a participant may decline to answer, a question may not apply because of skip logic<\/strong>, a measure may not be collected at a particular visit, a specimen may be unavailable, or a technical problem may occur. These situations have different implications. For example, a past\u201130\u2011day cannabis\u2011use item might be blank because a participant had never heard of cannabis (and was skipped past the question), because they declined to answer, because the measure was not collected at that visit, or because a technical failure interrupted data collection.\u00a0Datasets often use numeric placeholders such as 777 or 999 to represent refusal, \"do not know,\" or other non-substantive responses, although the meaning of these codes varies across studies and instruments and should always be verified in the documentation (ABCD Study, n.d.). These codes must be identified and interpreted appropriately; they are not substantive high scores.\r\n\r\n[caption id=\"attachment_410\" align=\"aligncenter\" width=\"1024\"]\"Flow Figure 3.2. Unavailable values can have different meanings.[\/caption]\r\n

    Why Missingness Matters<\/h3>\r\nMissing data can distort conclusions when available records differ systematically from unavailable records. Analysts describe missingness using three conceptual categories:\r\n