Smallholder maize farmers' experience and management of Fall Armyworm, seed access and crop protection: de-identified interview dataset, Siaya County, Kenya, 2025–2026

This dataset contains de-identified semi-structured interview data from 88 smallholder maize farmers in Siaya County, western Kenya, collected between June 2025 and January 2026. It documents farmers' experience of Fall Armyworm (Spodoptera frugiperda) infestation, the control methods they use and abandon, the criteria by which they judge a control technology, their indigenous and traditional practices, and their information sources and advisory relationships. Files README.md - full documentation of the dataset, its collection, de-identification, and terms of use FAW Siaya 2025 deidentified data.xlsx - participant-level dataset, 88 records × 75 variables FAW Siaya 2025 deidentified data.csv - the same data in plain-text CSV for software-independent reuse FAW Siaya 2025 codebook.xlsx - variable-level codebook, de-identification log and study notes FAW Siaya 2025 coding framework.xlsx - code definitions with inclusion and exclusion rules, anchor examples, salience, the analytic procedure, and the resulting design brief FAW Siaya 2025 thematic summaries.md - narrative summaries of the themes identified in the corpus, with illustrative quotations identified by anonymised participant code deidentify.py - the script that produced the de-identified dataset from the raw field export, provided so that every de-identification decision is inspectable and reproducible Collection. Face-to-face semi-structured interviews conducted in a language familiar to each participant, with responses entered into KoboToolbox during the session and complemented by field observation and field notes. Participants were recruited by purposive and snowball sampling with maximum-variation elements across gender, age, farm size, and experience, spanning six wards of Siaya County; recruitment continued to thematic saturation. De-identification. GPS coordinates, sub-location, village names, and platform metadata were removed; age, household size, and farming experience were banded; landholding was top-coded; and all open-text responses were screened for personal names, telephone numbers, and place names. The codebook records every action and its rationale and deidentify.py reproduces the procedure in full. Users must not attempt to re-identify participants. Appropriate use. Theme counts derived from these data indicate prominence within a purposive qualitative corpus. They are not prevalence estimates for Siaya County or any wider population. Ethics. Approved by the National Commission for Science, Technology, and Innovation (NACOSTI), Kenya, Licence No. NACOSTI/P/25/4175105. All participants gave informed consent prior to the interview. Contact. George Ariya, University of Eldoret - gariya@uoeld.ac.ke