Sheffield Alcohol and Network Dynamics
A six-wave residence-hall study of alcohol use, perceived norms, and social-network change. This page documents the data architecture, proxy/schema preview, and public-release preparation.
Build scope: study design, participant operations, REDCap/MySQL setup, de-identified linkage, QA checks, chapter datasets, proxy data, and public-release notes.
This page excludes: raw records, real identifiers, contact fields, free text, and real network ties.
Related: wave-by-wave network visualisations.
The study required the machinery around the analysis: recruitment, repeated survey delivery, identity separation, network-array preparation, and a reproducible pipeline that can run without exposing restricted data.
Recruited and onboarded 255 of 375 invited residents, then managed reminders, retention, incentives, and participant-facing support across six waves.
Maintained REDCap/MySQL infrastructure with study-specific survey links, email validation, anonymous enquiries, referral-code logic, and automated incentive workflows.
Separated contact details from behavioural responses and converted named important-peer nominations into pseudonymous IDs for analysis.
Built R/Make pipelines from raw exports to QA reports, longitudinal datasets, network arrays, model inputs, tables, figures, and provenance logs.
Prepared aligned behaviour arrays and directed adjacency matrices for network autocorrelation models and stochastic actor-oriented models.
Added proxy-data mode, generated dictionaries, privacy checks, checksums, validation notes, and this public study-book preview.
Sociocentric network data create an awkward requirement: students need to see names when they nominate peers, but the analysis files should not expose names, emails, or contact fields. SAND handled that with separate REDCap projects, an identity-manager key, and de-identified exports for the research workflow.
Participants selected names in the survey. Research exports carried coded identifiers, so the analysis could rebuild nominations without publishing names.
Baseline and follow-up data used different REDCap records. The linkage file connected those records while keeping names and emails outside the analysis repo.
Real edge lists, adjacency matrices, and anonymized network plots stay out of the public page. The site shows structure, schema, and proxy-mode documentation instead.
The study ran from September 2022 to October 2023. Later analyses use different subsets of the six waves.
| Wave | Fieldwork timing | Downstream use |
|---|---|---|
| Wave 1 | September 2022 baseline | Baseline covariates and pre-university alcohol measures |
| Wave 2 | October 2022 follow-up | First network measurement; Time 1 for selected longitudinal analyses; SAOM start |
| Wave 3 | November 2022 follow-up | Longitudinal preparation and quality assurance |
| Wave 4 | December 2022 follow-up | SAOM-aligned network wave |
| Wave 5 | March 2023 follow-up | Time 2 for selected longitudinal analyses; SAOM-aligned network wave |
| Wave 6 | October 2023 follow-up | Time 3 for selected longitudinal analyses; SAOM end |
Chapter 4 set out the workflow for consent, eligibility checks, identity management, survey invitations, coded nominations, and de-identified exports.
The Identity Manager held the key connecting names, emails, baseline record IDs, follow-up record IDs, and randomized study codes.
The research workflow used de-identified extracts. Contact details and name-labelled nominations stayed outside the analysis and public-site repositories.
These operational numbers give context for the infrastructure. They describe recruitment, retention, and response coverage, not substantive alcohol or network findings.
The study records alcohol use, alcohol-related consequences, perceived norms, and important-peer nominations across a bounded first-year residence cohort. The design gives the thesis both repeated measures and a changing peer network.
The study links individual alcohol trajectories to perceived norms and peer nominations. The thesis treats those as connected processes, not as separate survey modules.
The study measured alcohol outcomes, social norms, and directed important-peer nominations inside a fixed residence-hall boundary. Nominations begin at Wave 2, after students had time to meet one another.
Participants could nominate up to 10 people who had been important to them in the past month. Downstream scripts treat these as directed nominations.
| Feature | SAND design |
|---|---|
| Setting | One university-managed residence hall in South Yorkshire, UK |
| Invited population | 375 first-year residents |
| Enrolled sample | 255 students |
| Survey waves | Six waves, September 2022 to October 2023 |
| Network nominations | Up to 10 important-peer nominations per participant from Wave 2 onward |
| Behaviour outcome | AUDIT-C score, range 0-12 |
| Consequence measures | BYAACQ alcohol-consequence items, including a passing-out field used to construct a blackout-related outcome |
| Norm constructs | Descriptive and injunctive norm perceptions at global and important-peer levels |
| Main model families | Network autocorrelation models and stochastic actor-oriented models in the protected workflow |
| SAOM waves | Waves 2, 4, 5, and 6, with baseline covariates from Wave 1 |
The reproduction workflow covers the empirical chapters of the thesis. Each chapter receives prepared inputs from the previous stage rather than starting from a flat spreadsheet.
| Chapter | Role in the workflow | Primary data task |
|---|---|---|
| Chapter 4 | Data collection, coverage, and quality assurance | Build longitudinal measures, inspect wave coverage, and prepare network inputs. |
| Chapter 5 | Descriptive norms | Use prepared norm and alcohol measures for network autocorrelation models. |
| Chapter 6 | Injunctive norms and consequences | Track approval measures and alcohol-related consequence outcomes across analysis time points. |
| Chapter 7 | Social selection and influence | Align Waves 2, 4, 5, and 6 for stochastic actor-oriented modelling. |
In the public-release preparation workflow, real-data runs and proxy runs stay separate.
Approved researchers stage restricted exports locally. The files do not live in this site repo.
Synthetic inputs keep the same broad shape as the study data, so the workflow can run without real people or real network topology.
This page uses study-book text, schema summaries, and proxy-safe descriptions.
Scripts keep participant order stable across waves so behaviour arrays and network matrices refer to the same students.
The workflow derives peer means, perceived peer norms, and misperception scores from the nomination structure.
This page helps readers decide whether the protected SAND data would fit their question. It does not include records, raw identifiers, free-text responses, or real network files.
The dictionary groups repeated important-peer slots into patterns. It lists variable families, types, encodings, and waves.
| Variable | Type | Expected encoding | Description | Waves |
|---|---|---|---|---|
redcap_survey_identifier |
integer | Pseudonymous key | Within-study linkage key in protected data. Real values are not public. | 1-6 |
redcap_event_name |
character | Wave event label | Survey wave label used by the data-collection system. | 1-6 |
| Variable | Type | Expected encoding | Description | Waves |
|---|---|---|---|---|
age |
numeric | Years | Participant age at survey. | 1-6 |
sex |
integer | Binary analysis coding | Sex variable used in model covariates. | 1-6 |
ethnicity |
integer | Collapsed categorical code | Ethnicity category used in analysis data. | 1-6 |
majority_status |
integer | 0/1 composite | Composite majority-status covariate used where available. | 1-6 |
residence_cluster |
integer | Proxy-compatible grouping | Residence grouping used in proxy-compatible schema summaries. | 1-6 |
| Variable | Type | Expected range | Description | Waves |
|---|---|---|---|---|
q1 |
numeric | 0-4 | AUDIT-C item 1: drinking frequency. | 1-6 |
q2 |
numeric | 0-4 | AUDIT-C item 2: typical number of drinks per occasion. | 1-6 |
q3 |
numeric | 0-4 | AUDIT-C item 3: heavy episodic drinking frequency. | 1-6 |
audit_score |
numeric | 0-12 | AUDIT-C composite score calculated as q1 + q2 + q3. |
1-6 |
byaacq_6 |
numeric | Schema check range | BYAACQ-derived passing-out field used to construct a blackout-related outcome. | 1-6 |
The table names the nomination fields without listing nomination values.
| Variable | Type | Expected encoding | Description | Waves |
|---|---|---|---|---|
friend_number |
numeric | 0-10 | Number of important-peer nominations made by the participant. | 2-6; proxy schema may include Wave 1 |
which_friendid |
integer | 1-10 | Nomination slot index. | 2-6 |
nomination |
integer | Pseudonymous key | Identifier of the nominated important peer in protected data. Real values are not public. | 2-6 |
| Variable pattern | Type | Encoding note | Description | Waves |
|---|---|---|---|---|
inno1_self |
numeric | Approval scale | Participant's own approval of not drinking in social settings. | 1-6 |
inno2_self |
numeric | Approval scale | Participant's own approval of binge drinking. | 1-6 |
inno3_self |
numeric | Approval scale | Participant's own approval of drinking enough to pass out. | 1-6 |
deno{1,3,4}_friend_0 |
integer | Descriptive alcohol-use items | Perceived drinking frequency, heavy episodic drinking, and drunkenness for a typical resident. | 2-6; proxy schema may include Wave 1 |
inno{1,2,3}_friend_0 |
integer | Approval scale | Perceived approval of not drinking, binge drinking, and drinking enough to pass out by a typical resident. | 2-6; proxy schema may include Wave 1 |
deno{1,3,4}_friend_[1-10] |
integer | Descriptive alcohol-use items | Perceived drinking frequency, heavy episodic drinking, and drunkenness for each nominated important-peer slot. | 2-6; proxy schema may include Wave 1 |
inno{1,2,3}_friend_[1-10] |
integer | Approval scale | Perceived approval of not drinking, binge drinking, and drinking enough to pass out for each nominated important-peer slot. | 2-6; proxy schema may include Wave 1 |
| Variable family | Type | Description | Waves |
|---|---|---|---|
actual_*_peer |
numeric | Mean of the corresponding measure across nominated important peers. | 2-6; proxy schema may include Wave 1 |
deno*_peer |
numeric | Participant's aggregated perception of important-peer descriptive norms. | 2-6; proxy schema may include Wave 1 |
inno*_peer |
numeric | Participant's aggregated perception of important-peer injunctive norms. | 2-6; proxy schema may include Wave 1 |
misperception_*_peer |
numeric | Participant's perceived important-peer value minus the actual important-peer mean. | 2-6; proxy schema may include Wave 1 |
audit_score |
numeric | AUDIT-C composite calculated as q1 + q2 + q3. |
1-6 |