Delegating north: Device-linked cardinal reference and the topographic spatial option set in a Chinese mountain city

1. Overview Study title: Delegating north: Device-linked cardinal reference and the topographic spatial option set in a Chinese mountain city Design: Cross-sectional survey administered via the Wenjuanxing (问卷星) platform between July and December 2025. Population: Individuals currently residing or studying in Shiyan (十堰市), a prefecture-level city in northwestern Hubei Province, China, situated within the Qinba mountain system. Sample: N = 227 completed responses. After exclusion for geographic eligibility (Q5 registered city must be a Shiyan administrative unit; n retained = 146) and response validity (three participants providing uninformative placeholder text removed), the final analytical sample is N = 143. Instrument: A 27-item questionnaire (Q1–Q27) comprising six modules: (1) personal information and background (Q1–Q11), (2) spatial direction-word usage habits (Q12–Q16), (3) scenario-based spatial description tasks with visual vignettes (Q17–Q20), (4) dynamic open-ended route-giving tasks (Q21–Q22), (5) local landmarks and habitual spatial expressions (Q23–Q26), and (6) supplementary comments (Q27). The full instrument in Mandarin original with English translation is provided in Supplementary Appendix A of the published article. 2. File inventory Filename Description Format Codebook_Spatial_Reference_Survey.xlsx Variable codebook with four sheets: (1) variable definitions and coding rules for Q1–Q27, (2) frame-type classification for Q17–Q20 options, (3) GPS-delegation coding criteria for Q21–Q22, (4) exclusion criteria Excel (.xlsx) [dataset filename].csv De-identified individual-level survey responses (N = 143, post-exclusion). Available from the corresponding author upon reasonable request. CSV [analysis scripts] R / Python scripts reproducing all analyses reported in the article, including chi-square tests, confidence intervals, cross-tabulations, and discourse classification. R / Python 3. Key variables and coding Primary frame preference (Q12). Responses coded into three categories: cardinal (东南西北 "east-south-west-north"), relative (前后左右 "front-back-left-right"), and landmark (named environmental features such as buildings, mountains, rivers). This variable operationalises diagnostic criterion D1. Cardinal activation context (Q14). Multiple-select item. The GPS/map option (c) is the key variable for diagnostic criterion D2. Coded as selected (1) or not selected (0). Scenario frame choice (Q17–Q20). Each option carries a researcher-assigned frame-type label (cardinal, relative, landmark, topographic, or combination). Labels were assigned at instrument design by two authors independently applying Levinson (2003) criteria, with disagreements resolved by discussion. A third blind rater re-classified all 20 options, agreeing on 19 of 20 (95%); the single disagreement concerned Q19(e). Full classification in Sheet 2 of the codebook. GPS-primary delegation (Q21–Q22). Open-ended route descriptions coded as GPS-primary if the response directs the addressee to use a navigation application as the primary or sole strategy, with no substantive spatial content beyond the directive itself. GPS-primary responses further classified into three structural types: bare directive, hedged/face-managed, and idiomatically integrated. This variable operationalises diagnostic criterion D3. Full definitions in Sheet 3 of the codebook. Community cardinal awareness (Q24). Three-level ordinal variable: strong, moderate, weak (plus "don't know", excluded from cross-tabulation). Cross-tabulated with Q12 as an internal consistency check (§4.7). 4. Demographic variables used as covariates The following variables were tested for association with Q12 primary frame preference (§4.7). All yielded null results at this sample size (80% power to detect w ≥ 0.23): Daily dialect use (Q10): binary (yes/no). χ²(1) = 0.04, p = .84. Upbringing environment (Q6): dichotomised urban vs rural. χ²(1) = 1.34, p = .25. Education level (Q4): dichotomised postgraduate vs undergraduate. χ²(1) < 0.01, p > .95. Residence duration (Q7): long-term (born locally + 10+ years) vs short-term (<5 years). Borderline trend for GPS delegation rate (p = .078); see §4.7. 5. Exclusion criteria Two sequential exclusion criteria were applied (see Sheet 4 of the codebook): (1) geographic eligibility — only participants who selected a Shiyan administrative unit as their registered city in Q5 were retained (N: 227 → 146); (2) response validity — three participants who provided uninformative placeholder text for both open-ended route items (Q21 and Q22) were removed (N: 146 → 143). 6. Ethical approval and consent The study received ethical clearance from the lead author's institutional review board. All participants provided digital informed consent (Q1) before accessing the survey. Participation was voluntary and uncompensated. The informed consent specified that all collected data would be used for research purposes only, which restricts redistribution of individual-level responses to third-party repositories. 7. Missing data All closed-item questions (Q1–Q4, Q6–Q20, Q24) required a response; no missing values are present for these variables in the analytical sample. Open-ended items (Q5, Q21–Q27) permitted blank responses; however, Q21 and Q22 blanks triggered the response-validity exclusion criterion, so all 143 retained cases have non-empty Q21 and Q22 entries. Q23, Q25, Q26, and Q27 may contain blank entries.