Explore within-individual variation in MSM sexual activity over time using ARTnet partnership data

[smjenness]

Context

The recall-window argument (5-year vs 12-month) implies that a substantial fraction of self-identified MSM are inactive in any given year but active over a longer window. Steve Goodreau is skeptical that ~50% of MSM are inactive across a year, so the recall-window difference alone is unlikely to fully explain the Grey/Islek gap. ARTnet contains partnership data with recall periods that can bound this empirically.

This issue is also relevant to model-structure assumptions: ARTnet's parameterization currently has fixed one-time-partnership quintiles with no transitions between them, age and race-stratified main/casual partner degree distributions, and no individual-level random effects on degree. Activity cycling (or its absence) interacts with these assumptions.

Investigation tasks

• Inventory ARTnet's recall windows: partnership history, last-12-months, last-5-years, lifetime. Document which questions cover which windows.
• Estimate the fraction of self-identified MSM in ARTnet who report zero male partners in the past 12 months (NHANES-equivalent denominator) but nonzero in a longer window.
• Estimate the cross-sectional fraction with momentary degree zero (no current main, casual, or one-time partner). Compare to NHANES 1.75% and BRFSS 6.5% reference points where the comparison is meaningful.
• Assess whether the assumption of no quintile movement in one-time partnerships is consistent with the empirical ARTnet recall data, or whether it implies an implausibly stable activity classification.
• If activity cycling is non-trivial, sketch what ARTnet-derived parameters would drive an in/out cycling submodel (entry/exit rates from the active pool, possibly stratified by age and race).

Priority

Medium. Less time-critical than the literature audits but produces ARTnet-specific evidence on the recall-window question. Also produces inputs that may be useful to model structure work downstream.

Source

Plan: inst/popsize/msm_population_size_discussion.md, §Q3.