You are analyzing data from a registered single-arm longitudinal citizen science study (Efforia Study ID: 38316) titled "Porn Abstinence Impact: Nighttime Erection Trial" The study evaluates the physiological and psychological effects of 30 days of pornography abstinence in adult males. Participants served as their own controls across a structured 14-day baseline period followed by a 30-day abstinence intervention phase.

Baseline Structure — Critical for Analysis:

The baseline period consists of two distinct phases, each exactly 7 days, that are required but not necessarily sequential.

Baseline Week 1 (Days 1–7 from challenge start): Self-report only. Participants complete daily check-ins, the IIEF, AMS, PSS, and the "How You Feel About Yourself & Porn" survey while maintaining normal pornography habits. No device data is collected during this period.

Baseline Week 2 (7 days, device-dependent): Begins only after the Adam Sensor has been received and the first baseline sensor task (38333) is completed. Participants wear the Adam Sensor nightly and continue daily self-report check-ins. This phase may begin anywhere from immediately after Week 1 to several weeks later, depending on shipping.

The gap between Week 1 and Week 2 is a known, intentional design feature — do not treat it as missing data or dropout. Record the gap duration per participant and test whether gap length is associated with any outcome (it should not be, but verify).

Both phases must be completed before the 30-day abstinence period begins. Treatment start is individually anchored to commitment task completion, not a fixed calendar date.

Other key design features:

Single-arm — all inferential comparisons are within-person (pre vs. post)

Slips during abstinence are logged as data, not exclusion criteria

Participation is self-selected and participant-funded — account for potential selection bias toward men already motivated to reduce pornography use

PRIMARY OUTCOME MEASURES

For each measure, calculate the within-person change score (follow-up minus baseline). Where multiple follow-up assessments exist, use the final available assessment as the primary endpoint and treat intermediate assessments as secondary longitudinal data points.

Nocturnal Penile Tumescence (NPT) — Adam Sensor

Primary metric: Total Erection Time (minutes per night)

Secondary metrics (if available): frequency of erection events per night, average duration per event

Baseline window: All 7 nights of Adam Sensor data from Baseline Week 2 (after device arrival, before abstinence commitment)

Follow-up window: Final 7 days of the abstinence period (Days 23–30 of treatment)

Mid-study reading: Days 10–17 of treatment for longitudinal trajectory analysis

Do not include any sensor data from before the commitment task as treatment-phase data

International Index of Erectile Function (IIEF)

Use total IIEF score and domain subscores: Erectile Function, Orgasmic Function, Sexual Desire, Intercourse Satisfaction, Overall Satisfaction

Baseline: Week 1 administration

Follow-up: Final administration (post Day 30 of treatment)

Compute within-person change for total score and each domain

Aging Male Symptoms Scale (AMS)

Use total AMS score and subscales: Psychological, Somatic, Sexual

Baseline: Week 1 administration

Follow-up: Final administration

Note: Lower AMS scores = fewer symptoms = improvement. Ensure change scores are signed correctly (negative change = improvement)

Perceived Stress Scale (PSS) — Past Week Version

Use total PSS score

Baseline: Week 1 administration only — this is the anchor for stress stratification (see below)

Follow-up: Weekly administrations throughout treatment (repeated measure — analyze as a trajectory, not just endpoint)

Compute both endpoint change and slope across treatment weeks

Do not use Week 2 PSS as baseline — by Week 2, participants are aware of the upcoming commitment and may already be experiencing anticipatory stress or behavioral change

Daily Men's Health Check-In

Variables: mood, energy, libido, mental clarity, emotional resilience

Baseline Week 1 average: aggregate the 7 self-report days

Baseline Week 2 average: aggregate the 7 device-phase days separately

Report both as baseline reference points and note any Week 1 to Week 2 drift — systematic drift during the baseline gap may indicate behavioral anticipation effects

Treatment: weekly averages across the 30-day abstinence period

Model as a time series to capture trajectory shape

Weekly Porn Consumption Survey

Variables: frequency (days/week), estimated duration per session (minutes), slips during abstinence

Baseline: Average across both Week 1 and Week 2 baseline administrations

During treatment: week-by-week slip frequency and recurrence patterns

Compute: proportion with zero slips, 1–3 slips, 4+ slips

Test whether slip frequency moderates primary outcomes

STRATIFICATION VARIABLES

This analysis uses two primary stratification variables, applied independently and in combination.

STRATIFICATION 1 — TREATMENT EXPECTANCY

Definition: The degree to which a participant anticipated positive effects from pornography abstinence prior to beginning the intervention, captured in the Baseline Week 1 "Goals & Expectations" and "How You Feel About Yourself & Porn" tasks.

Operationalization:

Extract all expectancy-relevant items from the Goals & Expectations task (lesson ID: 90761) — items asking participants to rate anticipated improvement, expected difficulty, motivation for joining, and prior beliefs about pornography's effects

Construct a Treatment Expectancy Index (TEI) as a composite score. Sum or average Likert-scale items. For open-text responses, apply sentiment scoring (positive/negative/neutral) and convert to a numeric scale

Document TEI distribution, skewness, and floor/ceiling effects before splitting

Apply a median split: High Expectancy (HE) at or above the median, Low Expectancy (LE) below the median

Sensitivity analysis: run all analyses using tertile splits (Low / Medium / High) and report consistency

STRATIFICATION 2 — BASELINE PERCEIVED STRESS

Definition: A participant's self-reported stress level at study entry, captured by the PSS administered during Baseline Week 1. This stratification tests whether men entering the study under higher psychological stress show different physiological and behavioral responses to pornography abstinence.

Rationale: High baseline stress may amplify withdrawal-like symptoms during abstinence, suppress or enhance NPT recovery, and interact with pornography use patterns. Stress may also serve as a confounder if not accounted for — stressed men may use pornography more frequently at baseline, making their abstinence-period changes larger but potentially regression-to-mean driven.

Operationalization:

Use the Week 1 PSS total score as the stratification anchor

Apply PSS published clinical cut-points where available, or apply a median split: High Stress (HS) at or above the median, Low Stress (LS) below the median

Also test using PSS tertiles (Low / Moderate / High stress) as a sensitivity analysis

Document the PSS distribution at baseline — if the sample skews high, as expected in self-selected men seeking behavioral change, note this explicitly and consider whether tertiles better represent clinically meaningful distinctions than a median split

For the PSS outcome analysis: use Week 2 and treatment-phase PSS scores as longitudinal follow-up points, anchored to the Week 1 baseline value established for stratification

COMBINED STRATIFICATION — 2x2 EXPECTANCY x STRESS MATRIX

In addition to independent stratification, construct a 2x2 matrix crossing expectancy and stress groups. The four cells are HE-LS (High Expectancy, Low Stress), HE-HS (High Expectancy, High Stress), LE-LS (Low Expectancy, Low Stress), and LE-HS (Low Expectancy, High Stress).

Report sample sizes per cell — flag if any cell has n < 10 as underpowered for subgroup analysis

For each primary outcome, report mean change scores across all four cells

The most theoretically interesting contrasts are:

HE-LS vs. LE-HS — motivated, low-stress men vs. skeptical, high-stress men — tests whether psychological starting conditions predict differential physiological response

HE-HS vs. LE-HS — among high-stress men, does expectancy still matter?

HE-LS vs. HE-HS — among motivated men, does baseline stress moderate outcomes?

CORE ANALYSES

Analysis 1 — Primary Outcome Change by Expectancy Group

For each primary outcome measure:

Compute mean within-person change score for HE and LE groups separately

Report mean, SD, 95% CI, effect size (Cohen's d) for each group

Test between-group differences using independent samples t-test or Mann-Whitney U

Report p-values with appropriate caution given sample size and single-arm design

Analysis 2 — Primary Outcome Change by Stress Group

Replicate Analysis 1 exactly, substituting HS vs. LS as the grouping variable.

Pay particular attention to the AMS Psychological subscale and PSS trajectory — these are the measures most likely to show stress-group differences

Test whether HS participants show greater PSS reduction during abstinence (stress relief hypothesis) or greater PSS elevation (withdrawal/adjustment hypothesis)

Compare NPT change between HS and LS groups — test whether high-stress men show different nocturnal erection trajectories independent of expectancy

Analysis 3 — NPT Trajectory Analysis

Using the Adam Sensor data:

Plot mean nightly Total Erection Time across the study timeline for all four stratification groups (HE, LE, HS, LS) and all four 2x2 cells

Fit a mixed-effects model with time (week of study) as a within-person factor, expectancy group and stress group as between-person factors, an expectancy x stress interaction term, and time x expectancy, time x stress, and time x expectancy x stress interaction terms

Account for variable baseline start dates by anchoring time to individual treatment start dates

Account for the Week 1 / Week 2 gap by using device-on days for NPT calculations, not calendar days

Analysis 4 — Slip Analysis

Compare slip frequency between HE vs. LE groups and HS vs. LS groups

Test whether expectancy and/or stress predict slipping using logistic regression (any slip vs. no slip) and Poisson regression (slip count) with expectancy and stress as predictors

Test whether slips moderate primary outcomes — do participants who slipped show attenuated NPT improvement?

Examine whether high-stress participants slip more frequently and whether this accounts for any HS vs. LS outcome differences (stress → slipping → attenuated outcomes mediation)

Analysis 5 — Dose-Response: Baseline Porn Use vs. Outcomes

Stratify by baseline pornography consumption: Low (2 days/week or fewer), Moderate (3–5 days/week), High (6 or more days/week)

Within each stratum, compare outcome change scores

Cross-tabulate with both expectancy and stress groups — do high-consumption men cluster in particular expectancy/stress cells?

Flag regression to mean risk if high consumers show disproportionate improvement

Analysis 6 — Stress as a Mediator of NPT Change

Test whether reduction in PSS during abstinence mediates NPT improvement

Specifically test whether the pathway abstinence → stress reduction → NPT improvement holds as a mediating chain

Use bootstrapped mediation (5000 iterations), report indirect effect with 95% CI

Run separately within HE and LE groups — does the stress-mediation pathway operate equally across expectancy groups, or is it stronger in one?

Analysis 7 — Completers vs. Non-Completers

Define completers as participants who: completed the 30-day abstinence period, submitted 80% or more of daily check-ins across both baseline phases and treatment, and have baseline and follow-up data for at least one primary outcome

Compare baseline PSS, TEI, IIEF, AMS, and porn consumption between completers and non-completers

Test whether HE predicts completion using logistic regression

Test whether HS predicts dropout — does high stress make participants more likely to leave?

Conduct sensitivity analysis running all primary analyses on completers only vs. full available sample

SECONDARY & EXPLORATORY ANALYSES

Daily Users subgroup: isolate daily baseline porn users, compare their expectancy and stress profiles to non-daily users, and test whether they show larger NPT improvements

Time to first slip: Kaplan-Meier survival curves stratified by expectancy group and stress group separately

Self-report vs. objective concordance: compare IIEF erectile function subscale change with NPT change and test whether concordance is higher in HE or LS groups

Baseline Week 1 to Week 2 drift: test whether daily check-in scores (mood, libido, energy) change systematically during the gap between Week 1 and Week 2. If they do, this suggests anticipatory behavioral change before the commitment task and should be reported as a potential confound

Gap duration effect: test whether the length of the Week 1 to Week 2 gap (shipping delay) correlates with any outcome. It should not. If it does, investigate whether longer-gap participants differ behaviorally from shorter-gap participants

Responder analysis: define a meaningful responder as 20% or greater improvement in NPT combined with a 5-point or greater IIEF improvement. Report responder rates across all stratification groups and all four cells of the 2x2 matrix

Week-by-week PSS trajectory by stress group: test whether the HS group shows an early stress spike before later reduction and whether the LS group shows monotonic decline. Plot and compare trajectory shapes across both baseline weeks and all treatment weeks

STATISTICAL REPORTING STANDARDS

Report exact p-values throughout

Report effect sizes with 95% CI for all primary comparisons

Use two-tailed tests throughout

Apply Bonferroni correction for the family of primary outcome comparisons across both stratification variables — adjusted alpha accounts for 6 outcomes x 2 stratification variables = 12 primary comparisons (adjusted alpha = 0.0042)

Flag results significant at unadjusted but not Bonferroni-corrected thresholds as nominally significant

For all mixed models, report fixed effects, random effects structure, AIC/BIC

For the 2x2 matrix analyses, report all cell results even where underpowered — note power limitations explicitly rather than suppressing results

Present null and negative findings with equal prominence to positive findings

CONFOUND & BIAS ASSESSMENT

Flag and discuss the following:

Selection bias: self-selected, participant-funded sample. Assess whether baseline characteristics suggest a population predisposed to benefit

Expectancy-performance bias: HE participants may report better outcomes on subjective measures regardless of true physiological change. NPT is the critical objective check. If HE shows larger subjective but not objective gains, flag explicitly

Stress-confounding: high baseline stress may drive both higher baseline porn consumption and greater apparent improvement, creating spurious treatment effects. The HS vs. LS stratification is partly designed to detect this

Regression to the mean: high baseline consumers and high-stress participants may show apparent improvement due to regression. Cross-check NPT (objective) against self-report

Device compliance: flag participants with fewer than 4 nights of Week 2 baseline data or fewer than 10 nights of treatment data as potentially unreliable for NPT analysis

Week 1 / Week 2 gap behavioral contamination: participants aware of the upcoming abstinence commitment may reduce pornography use during the gap period even before officially committing. Test whether baseline Week 2 pornography consumption is lower than Week 1 at the group level. If so, document as a protocol-level limitation

Slip contamination: conduct per-protocol analysis excluding participants with more than 3 slips in the final 14 days of treatment

OUTPUT FORMAT

Present findings in the following structure:

Sample characteristics table: n overall and per stratification cell, baseline scores for all measures, TEI distribution, PSS distribution, 2x2 cell sizes

Primary outcomes table: mean change scores, SD, 95% CI, Cohen's d, p-value for each measure by expectancy group, stress group, and overall

2x2 matrix outcomes table: mean NPT change and IIEF change for all four cells

NPT trajectory figure: line graphs of mean weekly NPT by expectancy group and by stress group as two separate figures with error bars

PSS trajectory figure: week-by-week PSS means by stress group across baseline and treatment phases

Responder analysis table: meaningful responder rates across all stratification groups

Slip analysis summary: slip rates, time-to-first-slip curves stratified by both expectancy and stress, moderation results

Narrative interpretation covering: (a) whether objective NPT data support subjective self-report findings, (b) whether expectancy group differences suggest expectancy bias or genuine differential response, (c) whether stress group differences suggest stress as a moderator, mediator, or confounder of abstinence effects, (d) the pattern across the 2x2 matrix and which cell shows the strongest response and what this implies, and (e) implications for future RCT design including what should be stratified at randomization, what covariates should be collected, and what sample size would be needed to power the key comparisons based on observed effect sizes

LANGUAGE & FRAMING GUIDANCE

Present all findings as exploratory and hypothesis-generating, not confirmatory

Use associative language throughout — "abstinence was associated with" not "abstinence caused"

Explicitly acknowledge the absence of a control group in all interpretive statements

Where HE and LE groups show similar outcomes, frame as evidence against pure expectancy bias

Where HS and LS groups show similar outcomes, frame as evidence against stress as a primary moderator

Where findings are null or mixed, report with equal prominence to positive findings

Avoid language implying pornography is inherently harmful or beneficial — the study is explicitly neutral on this question