Moving 15 minute bin boundaries changed mean one hour relay rule gaps by up to 8.2 percentage points in six contact records

We tested how the position of time bin boundaries changes possible paths in six face-to-face contact records. The source was SocioPatterns data in Zenodo record 2540795. We used 334 complete one hour windows with at least ten observed people. One rule allowed one path step per bin. A second rule allowed any number of steps within each connected group in a bin. We compared their reachable ordered pair fractions. Four positions of 15 minute bin boundaries kept the same contacts, people, and observation windows. The range of the mean rule gap was 0.73 to 8.19 percentage points across records. The largest range was in the primary school. Its mean gap was 31.47 points at phase zero and 23.28 points at a 440 second phase. The mean gaps at phase zero were 5.42 to 46.95 points across the six records. Both rules bracketed the strict 20 second reference in all 10,020 tested cases. Bin width and observation length also changed the size of the gap. Shifted bins had partial bins at the window ends. Thus the phase test includes a change in the number of possible relay steps. The results describe possible paths under perfect transfer and unlimited waiting. They do not estimate actual information flow or infection. This planned sensitivity test extends two related SCI audits. It was selected for emphasis after we read those audits. The package contains code, aggregate results, figures, source hashes, and repeat instructions. Independent per-source checks and an exact numerical replay passed.

temporal networks · contact networks · reachability · time aggregation · bin phase · reproducible benchmark