Self-Rated English Proficiency and Beliefs About English Use Among Instructors of English-Medium and Chinese–english Bilingual Content Courses at Private Higher Education Institutions in Guangdong, China: A Cross-Sectional Survey With Measurement Diagnost

This cross-sectional online survey examined self-rated English proficiency, views on English use, and the measurement performance of the questionnaire among instructors teaching content courses through English-medium or Chinese–English bilingual instruction at private higher education institutions in Guangdong, China. Of 389 responses, 297 met the eligibility criteria and 260 remained after response-quality screening. The questionnaire included six general English-proficiency self-ratings (G1–G6), nine SRAE statements (S1–S9), four ELF-related importance statements (LO1–LO4), and six English-use questions (U1–U6). Mean SRAE was 3.334 (SD = 0.513), and agreement with the four importance statements ranged from 88.1% to 90.0%. The nine SRAE items showed high internal consistency (α = 0.907), and the ordinal one-factor model fit well, χ²(27) = 30.50, CFI = 0.999, RMSEA = 0.022. Across the 19 positively keyed scale items, 93.1% of respondents used only two adjacent numerical categories. In the bifactor model, the general factor accounted for 91.5% of common variance (categorical-data ωH = 0.957), while little reliable variance remained for the specific factors. A joint model of SRAE and the importance items was inadmissible (latent-variable correlation = 1.00004; polychoric HTMT = 1.002). Adjusted comparisons across the four cross-border-experience response combinations were imprecise, with all corresponding confidence intervals including zero. These findings support using more clearly differentiated questionnaires, verified eligibility procedures, and task-specific performance evidence alongside self-report when assessing faculty-development needs.