fsQCA Dataset for "Land Financialization and Urban Performance: A Configurational Analysis of Corporate Cities in South Korea"
Description
This dataset contains calibrated fuzzy-set membership scores for ten functional industrial districts (FIDs) in South Korea, used in fsQCA analysis of land financialization and urban performance. Includes variable formulas, calibration thresholds, and robustness test results. All raw data are derived from publicly available official statistics (KOSIS, KICOX, MOLIT, KIPO, Bank of Korea ECOS, KTDB).
Files
Steps to reproduce
The fuzzy-set membership scores reported in this dataset were produced through the following procedure. Step 1 — Raw data collection. Six composite indicators were constructed from publicly available official statistics: KOSIS (Statistics Korea) for employment, population, and regional industry data; KICOX (Korea Industrial Complex Corporation) for industrial complex performance and occupancy data; MOLIT (Ministry of Land, Infrastructure and Transport) for land price, land-use classification, and housing supply statistics; KIPO (Korea Intellectual Property Office) for patent grant data; Bank of Korea ECOS for regional economic accounts; and KTDB (Korea Transport Database) for travel time and commuting data. All indicators were averaged over a three-year reference period to reduce short-term volatility. Step 2 — Composite indicator construction. Each causal condition (A, B, C, D, E) and the outcome (Y) was operationalised as a weighted composite of two to three sub-indicators, standardised using z-scores prior to averaging. Full variable formulas are provided in Appendix A of the associated article and in the "Variables & Sources" sheet of this dataset. Step 3 — Calibration. Raw composite scores were calibrated into fuzzy-set membership values (0–1) using the direct method with a logistic transformation function: fs(xᵢ) = 1 / [1 + exp(−3(xᵢ − tᶜ))]. Three calibration anchors were set for each variable: full non-membership (0.05) at the 25th percentile, crossover (0.50) at the theoretically justified median threshold, and full membership (0.95) at the 75th percentile. Calibration thresholds and their qualitative rationale are documented in the "Calibration" sheet of this dataset and in Table 3 of the associated article. Step 4 — fsQCA analysis. Calibrated scores were entered into a truth table with consistency threshold 0.80 and frequency threshold 1. Necessary condition analysis used a 0.90 consistency threshold. Sufficiency analysis applied the Quine-McCluskey algorithm with directional expectations (positive for A, B, C, D; negative for E). Analysis was conducted using fsQCA 3.0 (Ragin & Davey, 2016) or equivalent set-theoretic software. Step 5 — Robustness checks. Six sensitivity tests were conducted as documented in the "Robustness Tests" sheet: consistency threshold variation (0.75/0.80/0.85), calibration crossover shift (±10%), leave-one-out case exclusion, alternative outcome specification, alternative E specification, and spatial unit sensitivity. All configurations remained stable across tests. Software. fsQCA 3.0 (available at www.compasss.org). Data preparation in Microsoft Excel. No proprietary data or software beyond standard official statistics databases were used.