Benjamini–Hochberg FDR-Corrected Chi-Square Tests for Associations Between IT Project Context and Business Analysis Practices (m = 792) (Ukraine 2020)
Description
This file contains the complete results of a retrospective Benjamini–Hochberg (BH) false discovery rate (FDR) correction applied to 792 Pearson's chi-square tests of independence. The tests examine statistical associations between IT project contextual attributes and business analysis practices, based on survey data from 328 IT professionals in Ukraine. All 792 chi-square tests are combined into a single BH family, since they all test variants of the overarching hypothesis: "contextual characteristics of an IT project are statistically associated with the selection of business analysis practices." The tests are organized into four subfamilies: Family A — Project context vs. requirements elicitation techniques (Q15). 180 tests: 12 context variables (project size, industrial sector, company type, company size, team distribution, experience, way of working, project category, elicitation responsibility, three types of elicitation sources) × 15 elicitation techniques. Collaborative games and Mind Mapping are excluded from testing due to insufficient observations (n < 10). Family B — Project context vs. specification and modelling techniques (Q19). 144 tests: 8 context variables (project size, industrial sector, company type, company size, team distribution, experience, way of working, project category) × 18 specification/modelling techniques. Goal models is excluded from testing due to insufficient observations (n = 4). Family C — Pairwise associations between specification and modelling techniques (Q19 × Q19). 153 tests: C(18, 2) = 153 unique technique pairs. Effect size measured as the phi coefficient (φ) for 2 × 2 contingency tables; absolute value of φ is reported in the V column. Family D — Project context vs. requirements documentation sections (Q18). 315 tests: 15 context variables (project size, industrial sector, company type, company size, system/service class, team distribution, experience, way of working, project category, templates usage, five types of documented requirements usage) × 21 documentation section types. BH procedure All 792 tests are ranked by raw p-value. For each rank k (k = 1, …, 792), the BH threshold equals k × 0.05 / 792. The adjusted p-value is computed as p_adj(k) = p_raw(k) × 792 / k, with monotonicity enforced from the bottom of the ranked list upward. An association is classified as significant (stable) if p_adj < 0.05. Results summary 154 out of 792 tests are significant after BH correction (p_adj < 0.05). All 154 significant associations have Cramér's V ≥ 0.10 (no trivial effects survive the correction). Effect size distribution: 11 high-effect (V ≥ 0.30), 143 moderate-effect (V = 0.10–0.29).By subfamily: A = 31, B = 16, C = 71, D = 36 significant associations. V range among significant associations: 0.145–0.413; mean = 0.215, median = 0.204.