Supporting data and code for "Temperature Response of Intrinsic Viscosity in Methacrylate and Hydrocarbon Viscosity Modifiers: A Harmonized Analysis of Published Data"

This record holds the data and analysis code underlying the manuscript “Temperature Response of Intrinsic Viscosity in Methacrylate and Hydrocarbon Viscosity Modifiers: A Harmonized Analysis of Published Data” (submitted for publication). The manuscript is not included here. The deposit contains: (1) Supplementary_Dataset_S1.xlsx, the harmonized evidence workbook of 32 worksheets, holding the source registry, the lineage audit, the intrinsic-viscosity observations with chemistry and architecture codes and molecular-weight provenance flags, the per-system temperature slopes, the direct-compatibility records with multiplicity weights, the chemistry-by-architecture coding table, the wax-window analysis, and all model outputs, with source-level provenance on every quantitative record; (2) Supplementary_Dataset_S2_Figure_Data.xlsx, giving the input and plotted values for every figure; (3) Supplementary_Software_S1.zip, the complete Python analysis pipeline with README.md, requirements.txt and MANIFEST.txt, which reads only the two workbooks and regenerates every reported table and figure; and (4) README.md and LICENSE.txt for the record. To reproduce: download all files into one folder, unzip Supplementary_Software_S1.zip, install the packages listed in requirements.txt, and run the following command from inside the Supplementary_Software_S1 folder: python reproduce_all.py --data ../Supplementary_Dataset_S1.xlsx --figdata ../Supplementary_Dataset_S2_Figure_Data.xlsx --out results The pipeline regenerates the one-stage mixed-effects model with publication and system random effects (methacrylate-minus-ethylene-propylene temperature interaction of +0.646 ln units per 100 °C, 95% CI 0.433 to 0.858), the system-slope regressions with wild cluster bootstrap inference (+0.775, p = 0.005), the wax cut-off sensitivity scan, the multiplicity-weighted cross-study compatibility model with chance-level comparisons, the domain-restricted ordinal phase model, and Figures 1–5 and S1. All resampling uses seed 20260918. A full run takes about one hour; adding --bootstrap 500 --reml-bootstrap 49 gives a fast check with identical point estimates. The study is a secondary analysis of published data; no new experiments were performed. Values in the exact-value core were taken from tables or explicit numerical statements in the cited sources, and figure-derived values are excluded from that core. Architecture is recorded as reported by the source authors and is not a measured quantity. The analysis code was prepared with the assistance of a large language model (Claude, Anthropic); all outputs and analytical decisions were checked by the author. Contact: Ziyad T. AL-Malki, Polymer Research Center, University of Basrah, Iraq ([e-mail]; ORCID 0000-0002-5137-6660).