Replication Data for: From Borders to Networks: The Changing Channels of Education Policy Diffusion in the American States

Purpose. The dataset supports full replication of all tables and figures in "From Borders to Networks: The Changing Channels of Education Policy Diffusion in the American States" (SPPQ-2026-0021). It allows verification of the discrete-time survival analysis showing that the dominant channel of education-policy diffusion shifted from geographic contiguity to policy-network proximity over the twentieth and twenty-first centuries. Nature. Two files. The source is the State Policy Innovation and Diffusion database, version 1.2 (Boehmke et al. 2020): 17,823 policy-state adoption records across 724 policies and 50 states, 1691–2017, with fields for policy, state, and adoption year. From it the build script derives the analysis file, a constructed state-policy-year survival panel of 13,112 observations and 27 variables, covering ten verified education policies. The constructed variables include the binary adoption event, a duration term, cumulative adoption, contiguity-based spatial lags (lagged and contemporaneous), network-exposure measures under three policy-pathway specifications, geographic and network centralities, region indicators, era markers, and a state-innovation activity index. Scope. The unit of analysis is the state-policy-year. Coverage is the fifty US states; Alaska and Hawaii enter as spatial isolates in the contiguity weights. The ten education policies span 1896 to 2014 in first adoption. A wider robustness file is generated on the fly from the same source, covering the 635 SPID policies with at least five adoptions (620,844 state-policy-years). All analysis variables are documented one to one in the codebook, and the panel is fully reconstructible from the source data by the included scripts, so no proprietary or restricted data are involved.