Accessible software modeling education for people with visual impairment: replication package for a systematic literature review

This is the replication package of a systematic literature review on accessible software modeling education for people with visual impairment. The review covers 35 primary studies published between 2004 and 2026, identified as A1 to A35, retrieved by automated search in five databases (ACM Digital Library, IEEE Xplore, Scopus, SpringerLink and the Brazilian Computer Society digital library), backward and forward snowballing, assisted manual search, and a cross-check against the corpora of two neighbouring systematic reviews. The package is distributed in two equivalent versions, one in English and one in Portuguese, each holding the same material: four spreadsheets, the analysis scripts and the three data figures of the article. The spreadsheets are the selection funnel stage by stage, the provenance of every study in the corpus, the corpus with its per-criterion quality scores and the extraction form, and the construction of the coding tab by tab, which classifies the 35 studies across four dimensions and 25 subdimensions and carries, for every mark, the literal passage of the study that supports it. The scripts reproduce the numbers reported in the article: the frequencies per subdimension, Fisher's exact test on 2x2 tables with Benjamini-Hochberg correction, the co-occurrence network and its null pairs, and the three figures. They are accompanied by the generators that produced the spreadsheets and by the code that translated them into English. At the root of the package, outside both versions, is the raw export of the automated search, with the 1,047 records retrieved from the five databases as the screening tool delivered them, and the status and criterion recorded for each. Every stage of the selection funnel can be traced back to it. Python 3 and the standard library, except the figure scripts, which need matplotlib, and one generator, which needs openpyxl. The full texts of the primary studies are not redistributed, being third-party material.