Contact pairs between departments repeat above a daily degree null at a 20-second threshold in two office records

We tested whether contact pairs between office departments repeat beyond daily partner counts. We used two public SocioPatterns records, labelled 2013 and 2015. Each file had ten nonempty days. A daily tie required at least 20 seconds of total contact. The null model kept each person's daily number of partners in every department fixed. It sampled each day independently with edge swaps. Across nine day transitions, the observed between-department repeat counts were 44 and 151. The null means were 25.21 and 98.37. Central 95% null ranges were 19 to 32 and 85 to 112. The observed counts were 1.75 and 1.54 times the null means. Removing the weekend transition kept the excess. At longer daily duration thresholds, the test had less freedom. All between-department blocks in the 2013 record were fixed under permitted swaps at 300 and 900 seconds. This gives no useful test of partner choice at those thresholds. Separate count code, small exact tests, longer runs, and an exact replay checked the analysis. Prior work already found office contact persistence. This contribution gives a scoped extension and a reusable test with stronger daily count constraints. It does not identify a social cause or establish a result for other workplaces. The package contains raw source files, provenance, code, all null draws, results, figures, and repeat instructions.

contact networks · temporal networks · office contacts · department structure · contact persistence · degree constraints · null model · reproducibility